AI Engineer working on the layer where LLMs meet real users — voice agents, RAG workflows, and agentic automation that has to behave when the demo is over.
I got here through backend: profiling PostgreSQL, fixing N+1 queries, stabilizing APIs. Then data, then product research, then applied GenAI. That path shapes how I build — retrieval design and prompt engineering on top, boring reliable engineering underneath.
The unglamorous parts are the ones I care about most: guardrails, fallbacks, evals, tracing, and knowing what every token costs.
Education
BCA — Data Science — Echelon Institute of Technology
AI / LLM
- LangChain
- LangGraph
- CrewAI
- LlamaIndex
- RAG
- Voice Agents
- Fine-tuning
- LLM Evals
Backend
- Python
- TypeScript
- FastAPI
- NestJS
- Node.js
- WebSockets
- Redis Queues
Data / Vector
- PostgreSQL
- MongoDB
- Redis
- Pinecone
- Qdrant
- pgvector
MLOps / Cloud
- AWS
- Docker
- Kubernetes
- GitHub Actions
- LangSmith
- RAGAS
- CI/CD
AI Engineer
Octify Technologies
Joined as an AI engineering intern, converted to full-time in six months. Owned LLM systems end to end — voice agents, retrieval, and agentic automation.
- Built voice-agent workflows for AI calling and sales coaching — prompt design, tool calling, schema-validated JSON outputs.
- Designed RAG pipelines with embeddings, vector search, and retrieval filters grounded in business and product context.
- Shipped agentic automation for lead generation, SEO monitoring, and WhatsApp/Twitter outreach — NestJS, async jobs, Redis state, retries, failure isolation.
- Hardened LLM flows for 10k+ requests a day: prompt versioning, guardrails, fallbacks, tracing, cost and latency tracking.
Spaze Palazo, Gurugram 
Mid-meeting Founding Engineer
Groovo AI
First engineer on a platform for growing teams that want to practice better sales conversations and produce marketing content more efficiently.
- Built the product from zero — AI conversation-practice flows and content pipelines for sales and marketing teams.
Product Research Intern
Filo Edtech
Turned scattered multi-channel user feedback into something the product team could actually act on.
- Built an internal feedback automation tool to collect, parse, categorize, and summarize feedback — research cycle time down 32%.
- Analyzed onboarding data, surfaced 3 high-friction points, helped lift retention by 5%.
All-hands at Filo Backend Developer
Panda Guys
Short backend stint building and maintaining REST APIs for client projects, working out of the HestaBit building.
HestaBit building, Delhi Backend Developer Intern
Get Now Solutions
First production backend work — where the performance instincts come from.
- Profiled PostgreSQL query paths in CRM APIs, fixed N+1 patterns, added indexes — p95 latency from ~800ms to under 200ms.
- Root-caused recurring REST API failures, cutting the backend bug rate by 40%.
Kind words
Jobbound
AI job-analysis assistant: fetches and deduplicates listings, scores fit against your resume with structured LLM outputs, finds gaps, and suggests tailored changes. Pre-LLM filtering, token-cost controls, zero-repeat checks, failure isolation.
- LLM Scoring
- Structured Outputs
- Next.js
Voice Agents for Sales Coaching
Voice-agent workflows for AI calling and coaching — prompt chains, tool calling, validated JSON outputs — serving 10k+ requests a day with guardrails, fallbacks, and full tracing.
- Voice Agents
- Tool Calling
- Guardrails
Agentic Growth Automation
Autonomous agents for lead generation, SEO monitoring, and WhatsApp/Twitter engagement — API tools, async jobs, Redis state, retries, and failure isolation in NestJS.
- AI Agents
- NestJS
- Redis
Groovo AI
Sales conversation practice and marketing content platform for growing teams — built from zero as founding engineer.
- LLM Product
- 0 → 1
Feedback Intelligence
Multi-channel feedback collection, parsing, categorization, and summarization — cut product research cycle time by 32%.
- NLP
- Automation