What I do well
Own backend end-to-end
Product problem to API, schema, migration, tests, deploy. At Adeptmind I own the GenAI QA evaluation platform this way, not ticket by ticket.
Dual-stack
Python (FastAPI, Django, SQLAlchemy) daily; TypeScript/JavaScript (Node.js, React) for 2 years across services and frontends.
Async and event-driven
Celery queues, WebSocket streaming, scheduled jobs, retries and idempotent workers in production.
LLM-backed product features
Tool calling, structured outputs, RAG, and grounding against real data - shipped and maintained for 3 years with LangChain, LiteLLM, and Mirascope.
Numbers from production
QA coverage
8% to 100% - automated customer call auditing with a GenAI evaluation platform.
Infra cost
40%+ reduction through backend and cloud optimization.
Latency
10% end-to-end cut in low-latency, event-driven AI workflows.
Current work
Role
Software Engineer, Adeptmind (Jan 2023 - present), remote from New Delhi.
What I build
Agentic QA evaluation: LLM orchestration, Django models, PostgreSQL, Celery, real-time streaming.
Education
B.E. ECE, NSUT Delhi (2018-2022).
Stack
Backend
Python, FastAPI, Django, Node.js, PostgreSQL, Neo4j, Redis
AI
LangChain, LiteLLM, Mirascope, RAG pipelines, vector databases, LLM evaluation
Infra
Docker, Kubernetes, AWS, Celery, WebSockets
Frontend
React, JavaScript/TypeScript