Hi, I’m Bishal Biswas, a Software Engineer based in Delhi, India, with 2+ years of experience building production AI applications using LLMs, RAG, and agent workflows. I specialize in AI application development, multi-agent orchestration, LLM integration, and backend systems using Python and AWS, with a focus on latency optimization, hallucination reduction, and production safety. I’m not focused on model training or ML research. I enjoy turning ideas into reliable, scalable solutions. My toolkit includes LangChain, LlamaIndex, RAG, tool calling, and HITL, and I’ve shipped end-to-end RAG pipelines, plan-and-execute patterns, and memory-aware architectures in fast-paced environments.

Bishal Biswas

Hi, I’m Bishal Biswas, a Software Engineer based in Delhi, India, with 2+ years of experience building production AI applications using LLMs, RAG, and agent workflows. I specialize in AI application development, multi-agent orchestration, LLM integration, and backend systems using Python and AWS, with a focus on latency optimization, hallucination reduction, and production safety. I’m not focused on model training or ML research. I enjoy turning ideas into reliable, scalable solutions. My toolkit includes LangChain, LlamaIndex, RAG, tool calling, and HITL, and I’ve shipped end-to-end RAG pipelines, plan-and-execute patterns, and memory-aware architectures in fast-paced environments.

Available to hire

Hi, I’m Bishal Biswas, a Software Engineer based in Delhi, India, with 2+ years of experience building production AI applications using LLMs, RAG, and agent workflows. I specialize in AI application development, multi-agent orchestration, LLM integration, and backend systems using Python and AWS, with a focus on latency optimization, hallucination reduction, and production safety. I’m not focused on model training or ML research.

I enjoy turning ideas into reliable, scalable solutions. My toolkit includes LangChain, LlamaIndex, RAG, tool calling, and HITL, and I’ve shipped end-to-end RAG pipelines, plan-and-execute patterns, and memory-aware architectures in fast-paced environments.

See more

Experience Level

Expert
Expert
Expert
Expert
Expert
Intermediate
Intermediate

Work Experience

Full-Stack Engineer at Build Champions
April 1, 2025 - Present
Evolved linear chains into stateful LangGraph DAGs, prioritizing complex reasoning over latency to hit 95% tool-selection accuracy. Boosted accuracy by 40% using GraphRAG and Hybrid Retrieval; refined RAGAS metrics to reduce hallucination patterns in 1k+ high-stakes records. Deployed hardened vLLM inference on SageMaker with PII masking and I/O guardrails, balancing safety against compute overhead. Integrated HITL and Plan-and-Execute patterns to mitigate autonomous trajectory failures, using LangSmith to monitor multi-turn memory.
Full-Stack Engineer at Independent Contract Work
November 1, 2024 - March 1, 2025
Designed multi-model LangGraph agents on AWS Bedrock, prioritizing semantic caching and DSPy-driven prompting to reduce hallucinations by 40%, balancing latency and accuracy. Built robust tool-calling workflows with strict JSON output validation and webhook-driven error handling, preferring deterministic routing for reliable production automation.
Full-Stack Engineer at Maxify (parent company - BePmeful Inc)
June 1, 2024 - November 1, 2024
Designed a high-concurrency RAG pipeline using LlamaIndex and Pinecone; implemented a hybrid Vector-Knowledge Graph structure to resolve semantic ambiguities in unstructured data, achieving a 25% lift in user discovery. Optimized 40k+ vector embedding workflows with Python async/await to reduce latency and improve retrieval accuracy.
Full-Stack Engineer at HBR Pvt Ltd
January 1, 2024 - May 1, 2024
Engineered RAG pipelines for CRM, reducing manual review by 60% (3k+ per month) through semantic search and document intelligence. Standardized data ingestion using AWS S3 and metadata tagging to improve factual grounding and lead scoring predictive accuracy.
Full-Stack Engineer at BePmeful Inc
August 1, 2023 - December 1, 2023
Implemented Memory-Driven system using LangChain and multi-turn memory for a scalable internal support framework handling 1K+ requests. Built Self-Reflecting Support Agents with context-aware chunking and prompt engineering to accelerate GenAI adoption.
Full-Stack at Tech Cooks L L P
May 1, 2021 - December 1, 2022
Dashboards and CRMs on cloud using React, Python, 2k+ users. Scalable REST APIs & containerized services, JWT and OAuth 2.0, and integrated third-party services like Stripe with webhook payment updates.

Education

Add your educational history here.

Qualifications

Bachelor of Computer Applications
January 1, 2019 - January 1, 2022

Industry Experience

Software & Internet