Headline

Software Engineer @ BOTmantra

About

I am a Backend Engineer specializing in building scalable APIs, LLM-powered retrieval systems, and automated data pipelines. At 2070 Health, I design high-performance systems that combine backend engineering, vector databases, and applied machine learning.Recently, I built a multi-vector semantic search engine using FastAPI, Qdrant, and OpenAI embeddings, improving candidate retrieval precision by 60% across 250K+ resumes. I also integrated GPT-4.1-mini based reranking pipelines, reducing inference latency by 94% and boosting relevance scores through structured prompt engineering.My experience spans:• Designing LLM/RAG retrieval pipelines• Building versioned MLflow-Qdrant evaluation systems• Implementing ETL/ELT workflows (Airbyte → Redshift → DBT → Superset)• Creating secure AWS Lambda APIs with presigned S3 URLs• Developing Slack/Outlook/Teams scheduling automation that cut HR workload by 75%I enjoy solving hard engineering problems at the intersection of backend systems, data engineering, and AI.Always open to collaborating, learning, and building impactful products.

Location

Bengaluru

AI stack

claude code

claude

langchain

langgraph

Work experience

Software Engineer at BOT mantra, Feb 2026 - Present · 10 mos, Bengaluru, Karnataka, India

Software Development Engineer 1 at 2070 Health, May 2024 - Jan 2026 · 1 yr 9 mos, Gurugram, Haryana, India, • Built a multi-vector semantic search engine (FastAPI + Qdrant + OpenAI embeddings) improving candidate discovery speed by 60% across 250K+ resumes. • Created NLQ-based parameter extraction, UUID tagging, and Postgres query logging, improving search interpretability and audit traceability by 40%. • Enhanced Qdrant with custom weighted-fusion modules and reranking logic, boosting top-N precision by 30%. • Delivered a secure resume-retrieval API using AWS Lambda, API Gateway, Authorizer, and presigned S3 URLs, enabling scalable, permissioned file access. • Automated LinkedIn activity ingestion with Sheets + PhantomBuster + MongoDB/Postgres upserts, cutting ops work by 70% and achieving 0 duplicate processing. • Migrated n8n automations into Python ELT pipelines (Zoho → Redshift), increasing reliability and reducing deployment overhead by 65%. • Built a PageSpeed monitoring pipeline (Airbyte → Redshift → DBT → Superset) reaching 90–95% automation and reducing manual reporting by 95%.

Graduate Engineer Trainee at Ericsson, Jan 2024 - May 2024 · 5 mos, India

Education

Bachelor of Technology - BTech, Computer Science, sikkim manipal institute of technology