Fullstack AI Engineer crafting production LangGraph multi-agent DAGs, PostgreSQL pgvector hybrid search, Python PySpark data ETL pipelines, and Next.js TypeScript web applications.
Core production systems built as an individual contributor in cross-functional engineering teams.
Enterprise Multi-Agent Workflow Engine
Stateful asynchronous Python multi-agent DAG engine with Redis checkpointing, human-in-the-loop validation, and streaming trace telemetry.
Multi-Tenant Hybrid Vector & Graph Retrieval
Scalable multi-tenant hybrid search engine using pgvector HNSW indexing, LlamaIndex parent-document retrieval, BGE-Reranker, and chunk-level RBAC.
Custom telemetry gateways, PySpark ETL pipelines, and specialized AI developer tools.
Real-Time Proxy & Cost Optimization Plane
High-throughput Python AsyncIO reverse proxy for real-time LLM telemetry, automated PII scrubbing, fallback provider routing, and token cost budgeting.
High-Throughput PySpark & DuckDB Store
Scalable Python data engineering pipeline using PySpark, DuckDB, and Parquet for automated data cleaning, feature extraction, and real-time model ingestion.
Stateful asynchronous workflow breakdown with Redis state checkpoints and human-in-the-loop intervention gates.
Multi-tenant dense HNSW vector embeddings combined with BM25 sparse search and Cohere cross-encoder reranking.
High-throughput Python AsyncIO reverse proxy delivering sub-6ms PII regex scrubbing, token cost budgeting, and Langfuse tracing.
High-throughput PySpark batch processing combined with DuckDB zero-copy in-memory querying for feature extraction.
NeuralScale Labs · Full-time
Engineered multi-agent LangGraph workflow pipelines, PostgreSQL pgvector hybrid retrieval engines, and Next.js TypeScript web applications.
Apex Global Tech · Full-time
Developed high-concurrency Python FastAPI microservices, PostgreSQL databases, and React/Next.js frontend user interfaces.