Open to the right Full-Stack AI opportunity
Nikhil Rajput · Full-Stack AI Engineer
I build AI products that move from conversation to production.
I turn business requirements into reliable agents, automation systems, full-stack platforms, data pipelines, and SaaS products — then help them survive real users, real integrations, and real operational constraints.
- Doing this
- 2+ years, hands on
- Based in
- Madhya Pradesh, India
- Scope
- Discovery → production
Fig. 1One request, all the way through
the boring parts are the product
100+development applications captured dailydata acquisition · Australia
20+council portals tracked overnightdata acquisition · Australia
250+property listings ingested monthlyPropTech SaaS · South Africa
<500msto first audio on live PSTN callsvoice AI · India
16+safety guardrails on every model calllegal intake · Australia
1,000+students managed per academic sessioneducational ERP · India
What I bring
The intelligence layer is the easy half.Everything it touches is the job.
My best work happens in the overlap — where AI, full-stack engineering, integrations and somebody's actual working day all have to agree with each other.
I’m a Full-Stack AI Engineer with 2+ years of hands-on experience building AI-powered products, automation systems, full-stack SaaS platforms, data pipelines, and production integrations for international clients.
I work from the first confused conversation through architecture, build, testing, deployment and the long tail of changes that follow. I'd rather carry that line end to end than hand off half of it.
Read my complete résuméI start where the job actually happens
Not the one in the process document. I map the people, the spreadsheets, the WhatsApp group where the real decisions happen, and the edge cases nobody mentions until week three.
I own the connected build
Interfaces, APIs, databases, agents, workers, integrations, tests and the deploy pipeline. One person holding all of it means fewer seams for things to fall through.
I design for the day it misbehaves
Structured outputs, validation, bounded tools, retries, fallbacks, logging and a clean human handoff. Not features bolted on later — the shape of the thing from the start.
What I've shipped
Shipped systems, with theengineering story attached.
Client delivery across AI, automation, data acquisition and SaaS — then my own products, where I get to make the architecture calls and live with them.
Client systems · 05 in production
Delivered for teams in South Africa, Australia and India — each one replacing work that used to happen by hand, in a spreadsheet, or not at all.
Multi-Tenant AI PropTech SaaS
Listings, tenants, outreach and syndication running as one multi-tenant system.
A full-stack property platform where acquisition, tenant operations, owner outreach and social distribution all share one isolated-per-tenant data layer, with AI agents that act on that data rather than just describe it.
- Data harvesting
Configured session-managed Playwright scraper tasks to aggregate, sanitise and ingest 250+ rental listings a month from diverse portals into an isolated multi-tenant database.
- Autonomous outreach
Built a communications engine that sends hyper-personalised responses to property owners the moment a tenant expresses interest, and extracts intent from the replies that come back.
- Social syndication
Orchestrated n8n workflows that generate AI-written listing content and distribute it to Facebook Marketplace, Facebook Pages and Instagram, cutting manual data entry by over 90%.
- Action-oriented agents
Shipped a WhatsApp RAG bot and an in-app agent that execute real database commands, so a user can list a property or raise a service request entirely through natural language.
Node.jsExpress.jsTypeORMPostgreSQLChromaDBLangChainRedisSocket.IOPlaywrightn8nWhatsAppFacebook Graph APIInstagramDocker
AI Legal Intake & Safety Infrastructure
An SMS intake agent that qualifies claimants without ever giving legal advice.
A configurable intake platform that collects facts, scores follow-up readiness and escalates to a lawyer, with safety gates on both sides of every model call.
- AI intake and qualification
Designed an SMS-based intake platform on FastAPI and Twilio with configurable conversation flows, structured fact extraction, lead scoring and human escalation before any lawyer review.
- Safety and guardrails
Engineered inbound and outbound safety gates with crisis detection, restricted-response rules and output validation across 16+ documented guardrails, routing critical cases straight to a human.
- Deterministic orchestration
Built a conversation engine that combines rule-based classification, configurable intake questions, explicit scoring logic and fallback model handling inside one controlled flow.
- Compliant by construction
Kept every model call inside Australian-region inference for data residency, with an append-only audit log and a separate guardrail-event trail for review.
PythonFastAPITwilioAWS BedrockClaudePostgreSQLAWS App RunnerSQSCloudFrontHubSpotTerraformNext.js
Conversational Voice AI & Marketing Agents
Live PSTN calls under two seconds, handed to a human the moment intent demands it.
A real-time voice stack that holds a natural conversation, carries its context into WhatsApp afterwards, and absorbs the repetitive front half of a marketing team's day.
- Low-latency audio streaming
Built a voice telephony stack on FastAPI and webhooks against an explicit latency budget — roughly 150ms speech-to-text, 400-600ms model and 200-400ms to first audio — with time-to-first-audio instrumented on every turn.
- Contextual continuity
Logged full conversation state to a central store so automated onboarding — delivery partners and education enquiries alike — moves from a phone call into a WhatsApp RAG assistant without losing the thread.
- Workforce multiplier
Automated initial lead scoring and high-volume enquiry handling, scaling daily capacity to absorb the work of four to five marketing executives.
- Intelligent handoff
Embedded LLM function-calling triggers that read speech intent in real time and execute an instant warm transfer to a live agent.
PythonFastAPIPlivoGeminiGroqClaudeSarvamWhisperWebSocketsWhatsAppPostgreSQLDockerGCP Cloud Run
Enterprise Data Acquisition Platform
Twenty council portals watched overnight, turned into enriched CRM leads by morning.
A distributed acquisition pipeline that tracks public planning portals, reads the documents behind each application, enriches the record and routes it to the sales team automatically.
- Distributed scraping engine
Designed resilient Playwright workers that continuously track 20+ Australian local council portals, harvesting and indexing 100+ new development applications daily.
- LLM parsing pipeline
Wrote asynchronous entity extraction on AWS Bedrock that pulls builders, applicant contacts and project scope out of complex, unstructured planning documents.
- Automated enrichment
Embedded real-time lookups through the ABR ABN Lookup API and RP Data, feeding official corporate contacts and property analytics into every extracted lead.
- CRM sync and orchestration
Built a React monitoring dashboard wired to Monday.com, triggering instant client email workflows for new leads and removing manual routing and data entry by 95%.
PythonPlaywrightAWS BedrockAWS TextractReact.jsPostgreSQLMonday.comABR APIRP DataEC2S3Docker
Construction Slab Layout & Take-Off Platform
Draw the slab, place the pods, get the measurements the site orders against.
A construction planning tool that takes concrete raft-slab boundaries, works out the void-former pod layout across every slab on the plan, and feeds the resulting measurements straight into the engineering calculation sheet.
- Slab take-off
Parses multi-slab plans into boundary geometry with Shapely and normalises each slab so the pod grid lands on the correct setout however the plan was drawn.
- Pod placement
Built the placement pass that fills each slab with full pods, then works the perimeter with fractional pods, checking containment and overlap against the real boundary rather than a bounding box.
- Measurement and quantities
Wired the results into the engineering Google Sheet — slab thickness, pod height, slab area and rebates written in, calculated quantities read back — so ordering and measurement stay in one place.
- Delivery
Shipped it as a Flask service with MongoDB, JWT and service-account auth behind a React workspace, with Plotly setout drawings for each candidate layout.
PythonFlaskShapelyNumPyPlotlyMongoDBGoogle Sheets APIGoogle Drive APIJWTReactViteGunicorn
Integration & automation
The tools a business already runs, finally talking to each other.
Any tool with an API
CRM, inbox, WhatsApp, voice, LinkedIn, Drive and Sheets are the ones already wired — the orchestration layer takes anything with an API, deterministic where it matters, with agents only on the language-shaped steps.
Cross-Tool Workflow Automation Suite
One automation layer wired across CRM, inbox, WhatsApp, voice, LinkedIn and Sheets.An n8n-centred automation layer that joins the tools a business already runs — HubSpot, Gmail, Brevo, WhatsApp, voice calling, LinkedIn, Drive and Google Sheets — and lets LLM agents handle only the steps where the input stops being structured.
n8nHubSpotWhatsApp Business APIBrevoGmailGoogle SheetsGoogle DriveLinkedInVoice / calling APIsNode.jsChrome ExtensionsWebhooksLLM agents
CRM and pipeline
Two-way HubSpot sync across contacts, deals, notes and activity, with leads captured from web forms, LinkedIn and inbound calls deduplicated and dropped into the right pipeline stage without anyone retyping them.
Messaging and voice
WhatsApp Business, Brevo transactional email and voice-flow calling behind one workflow, so a contact is reached on whichever channel they actually answer — and every reply, call outcome and opt-out lands back on the CRM record.
Outreach and sourcing
A LinkedIn outreach extension whose calls all execute in the user's own browser rather than from a server IP, with profile-aware messages, queueing, sequenced follow-ups, controlled delays and A/B variants.
Documents and records
An invoice pipeline across webhook, Gmail and Drive triggers, where LLM extraction is fenced by schema validation, duplicate checks and a retry queue before anything writes to Sheets or the CRM.
Agents and control
Bounded LLM agents for classification, personalisation and extraction, with deterministic routing around them, alerting on every failed branch, and a human review step wherever an action is hard to undo.
My own products
Where I make the architecture calls and live with them.
03 AI products
Independent product work — evaluation, data boundaries, async systems and the interface all move together, at my own risk.
WebPilot AI
Teach a browser workflow once, then replay it with zero model calls — and heal it when the site changes.An autonomous web operations platform that turns a plain-language goal into an approval-ready typed workflow, learns it against a live browser, freezes it as an immutable spec, and replays it deterministically until the page underneath moves.
- Built a two-mode engine: the first run lets Gemini plan and drive a real browser step by step, and every healthy run after that replays a frozen, typed spec through Playwright with zero reasoning calls.
- Made self-healing a pipeline rather than a retry — diagnose the failure from DOM and screenshot evidence, patch it, replay in a sandbox, have a second independent model verify the patch, promote a new immutable version, then resume the same run.
Next.jsNestJSFastifyTypeScriptPlaywrightGeminiGoogle ADKVertex AIPostgreSQLCloud RunCloud TasksPub/SubCloud SQLSlack APIDocker
AI Career & Job Intelligence Platform
An AI-first operating system for the entire job search, not one more resume tool.Career profiles, resumes, portfolios, job intelligence, matching, applications and interview preparation live inside one connected product architecture.
- Building scheduled connectors for LinkedIn, Indeed, Naukri and Wellfound, with parsing, normalisation, duplicate detection, skill extraction and Weaviate-backed vector matching.
- Running parsing, scraping, matching, resume generation and notifications on Redis and Taskiq workers, so nothing slow ever blocks the product experience.
- Routing every model call through a LiteLLM gateway with LangGraph workflows and LangSmith traces, plus a pluggable voice adapter for spoken interview practice.
Next.js 16React 19FastAPIPostgreSQLWeaviateRedisTaskiqLangGraphLiteLLMLangSmithSarvamDocker
Financial Analytics AI Platform
Static bank statements become a queryable, auditable financial asset.A finance platform that parses long PDF statements into clean transactions, then answers questions about them with retrieval grounded in the user's own ledger.
- High-volume ingestion: Built a Python and FastAPI extraction pipeline that parses and cleans unstructured transaction data out of 100+ page PDF bank statements.
- Hybrid retrieval: Structured a RAG layer combining LangChain, pgvector semantic search and generated SQL so questions resolve against real rows rather than a summary.
- Interactive visualisation: Built a React dashboard that categorises and charts expenses and cash-flow trends across 1,000+ historical transactions.
- Automated auditing: Turned static documents into queryable assets, cutting manual auditing and transaction-filtering turnaround by around 85%.
PythonFastAPIReactPostgreSQLpgvectorLangChainLangGraphSQLAlchemy 2.0asyncpgDocker
Additional products
More proof, across more categories.
05 more systems
A compact view of the wider full-stack, integration, mobile, language and geospatial systems I've worked on.
Omnichannel MarTech SaaS
Publishing and paid campaigns across five networks from a single console.A centralised product for secure channel connections, personalised publishing, and paid advertising operations across the major social platforms.
- Multi-platform automation: Designed a full-stack system with OAuth 2.0 flows across 5+ platforms — Meta, TikTok, YouTube, LinkedIn and X — enabling automated personalised publishing to client accounts at scale.
- Paid advertising orchestration: Built a centralised campaign layer spanning Meta, TikTok, Google, LinkedIn and X Ads, with objective mapping and consolidated spend reconciliation in one place.
Next.jsReact.jsNode.jsExpress.jsPostgreSQLOAuth 2.0Meta AdsTikTok AdsGoogle AdsLinkedIn AdsX Ads
Enterprise Educational ERP
The full student lifecycle for a thousand students an intake, in one system.An ERP built around the daily workflows of students, faculty and administrators, from admission through to placement.
- Institutional architecture: Launched an ERP managing the end-to-end lifecycle of 1,000+ students per academic session across admissions, interviews, attendance and placement tracking.
- Mobile-native verification: Replaced legacy attendance hardware with a cross-platform, GPS-geofenced QR scanning layer that keeps check-in and check-out authentic.
- Administrative suite: Built a high-throughput portal supporting 20+ faculty and staff in digitising day-to-day academic operations and certification.
React.jsNode.jsExpress.jsMySQLWhatsApp TemplatesPhonePeQRGPS
Real Estate CRM Bridge
A third-party CRM normalised into a clean multi-tenant schema, live on every device.Property listings, contacts and enquiries stay synchronised between an external CRM, operational dashboards and a mobile app.
- Built a FastAPI sync layer that consumes listings, properties, contacts and enquiries from a third-party CRM and normalises its nested JSON into a clean relational schema.
- Designed it multi-tenant from the start, so onboarding a new agency is a sync script rather than a deployment, with real-time Firebase updates pushed to every client.
- Delivered an Expo mobile workflow for on-the-go property management, client communication and listing updates.
PythonFastAPIPostgreSQLReactReact NativeExpoFirebaseREST APIs
Field Interview Transcription Pipeline
Hour-long Hindi field interviews become schema-enforced agricultural records.An audio-to-structured-data pipeline that converts long-form farmer conversations into data ready for downstream analysis.
- Combined Google Cloud Speech-to-Text with Gemini to turn long Hindi field recordings into structured agricultural records.
- Enforced crop, soil and farming-practice fields through Pydantic schemas, with retry and human review paths when a field cannot be filled confidently.
PythonFastAPIGoogle Cloud STTGeminiPydantic
Precision Agriculture Platform
Nine satellite analysis layers for any field you can draw on a map.A geospatial platform that turns remote-sensing data into zone-wise farm analysis and temporal NPK trends.
- Generated nine satellite analysis maps per field — vegetation, moisture, terrain, soil and nutrients — from Sentinel Hub, with ISRO VEDAS as an optional India reference overlay.
- Built an OpenLayers workspace where drawing a farm boundary drives the whole analysis, plus NPK time-series and prescription maps for that field.
PythonFastAPIReactOpenLayersSentinel Hub APIISRO VEDASPostgreSQL
Every system above shipped. Ask me what broke first — that story is usually more useful.
How I work
Useful AI startslong before the model call.
The model is one component. The workflow, the data it's allowed to see, the controls around it and the recovery path decide whether anyone keeps using the thing in month two.
four habits, in order
- 01no architecture on day one
Sit with the process before the stack
Before choosing a model or a framework, I map who does the job today, what they actually touch, where the handoffs break, and what the business is really paying for.
- 02the leash is the design
Draw the line between rules and reasoning
Money, eligibility, compliance and anything irreversible stay deterministic. The model gets the messy, language-shaped middle — and a short leash of allowed tools.
- 03one problem, not seven
Build the full path, not the clever part
Interfaces, APIs, workers, queues, databases, integrations, logging and deploys are one problem. A great agent behind a broken webhook is still a broken product.
- 04assumptions lose
Ship it into the working week
I test the ugly paths, release in slices, watch what production does to my assumptions, then change the system with the people using it.
Technical range
Grouped by the part of the systemthey hold up.
I pick the stack around the workflow, the deployment environment and the reliability bar — not around whatever I used last time.
AI & GenAI
LLM APIsAI AgentsRAGStructured OutputsEmbeddingsVector SearchGuardrailsTool CallingLangChainLangGraphLangSmithLiteLLMOpenAIGeminiAnthropicAWS Bedrock
Backend
PythonFastAPINode.jsExpress.jsREST APIsWebSocketsWebhooksAsync ProcessingBackground WorkersAuthentication
Frontend
React.jsNext.jsTypeScriptJavaScriptReact NativeTailwind CSSResponsive DashboardsOpenLayers
Automation
n8nPlaywrightBrowser AutomationWorkflow OrchestrationWhatsApp AutomationVoice AITwilioData PipelinesWeb Scraping
Data
PostgreSQLMySQLMongoDBChromaDBWeaviatepgvectorRedisSQLAlchemy
Cloud & DevOps
AWS EC2S3App RunnerCloudFrontSQSLambdaTextractGCP Cloud RunDockerGitHub ActionsCI/CDTerraformBitbucketVercel
Integrations
HubSpotMonday.comMetaTikTokYouTubeLinkedInXWhatsAppPayment GatewaysABR APIRP DataBrevoFirebaseSentinel Hub
How the responsibility grew
Quality first. Then software.Now complete AI systems.
It explains how I work now: I still hunt edge cases like a tester, build like an engineer, and keep the product in view the entire time.
Singaji Software SolutionsJuly 2024 — present
Three roles in two years, each one further up the stack and further into ownership.
03
Nov 2025 - Present
Full-Stack AI Engineer
I lead the development of AI-powered automation and full-stack systems across architecture, LLM workflows, backend services, integrations, testing, deployment, and production iteration.
- LLM applications, AI agents, RAG pipelines, structured extraction, tool calling, guardrails, and human-in-the-loop workflows
- Production systems with Python, FastAPI, Node.js, PostgreSQL, n8n, Playwright, Redis, AWS, and third-party APIs
02
Nov 2024 - Nov 2025
Software Engineer
I developed backend and full-stack applications for international clients across Australia, South Africa, and other markets.
- Node.js, Express.js, Python, FastAPI, React, Next.js, PostgreSQL, and MySQL
- REST APIs, webhooks, integrations, dashboards, scraping pipelines, data workflows, deployment, and production support
01
Jul 2024 - Oct 2024
QA Tester + Developer
I started close to product quality: testing real workflows while contributing to development, debugging, and application reliability.
- Functional, API, regression, and workflow testing
- Defect reproduction, edge-case validation, fix verification, and collaboration with developers
Where I fit next
Give me a real problem,not an isolated demo.
I'm looking for a team where engineering ownership, AI capability and actual business context sit in the same room.
Agents that understand, decide, and act without going off the rails
Full-Stack AI Engineer · GenAI Engineer · AI Automation Engineer
The complete product wrapped around the intelligent part
Full-Stack Developer · AI Product Engineer · Applied AI Engineer
Business friction on Monday, something shipped by the end of the quarter
AI Solutions Engineer · Automation Engineer · Product-minded Engineer
What a team gets
I don't stop at“the AI part.”
I can sit with the people doing the job, turn a vague complaint into a decision, and keep the engineering pointed at the outcome instead of the ticket.
System boundaries and schema design through to APIs, agents, dashboards, automation, integrations and the deploy. I go all the way down.
I know when plain code beats a model, where AI genuinely earns its cost, and where validation, fallback, observability or a human has to be in the loop.
Start a conversation
Got a process thateats everyone's week?
I'm in Madhya Pradesh, India, open to Full-Stack AI, GenAI, full-stack, automation and AI solutions roles, and happy to work with teams in other timezones. Tell me what your team is trying to make faster, cheaper or less painful.
