Sayam Monga

About Sayam Monga

I am a Software Engineer specializing in ML Systems and AI Infrastructure with 2+ years of experience bridging high-performance systems engineering (C++) and modern Generative AI architectures (Python).My core expertise lies in optimizing AI at the algorithmic and hardware levels. Recently, I architected a custom in-memory AI architecture that utilizes hashmap-based data structures to train models directly on CPUs—completely bypassing standard GPU matrix-multiplication bottlenecks and drastically reducing compute costs. I also built a high-performance C core library implementing complex mathematical algorithms for high-throughput execution and strict memory safety.On the applied AI side, I have engineered end-to-end RAG pipelines, Agentic workflows (LangGraph with MCP integration), and dynamic MoE (Mixture of Experts) routing layers to reduce token costs and latency. Additionally, I have implemented Responsible AI (RAI) guardrails to ensure output validation and safety. I am passionate about writing scalable code and building robust data pipelines to deploy cutting-edge AI models into production.

Experience

1. Software Engineer

   Datopic Technologies

   Jul 2024 — Present · Mohali, IN

   Awarded ‘Shining Star of the Year’ (2024) for architecting a high-performance C core library,implementing complex lattice-based mathematical algorithms (Dilithium PQC) to achieve high-throughput execution and strict memory safety.•Architect an In-Memory CPU Training Framework, utilizing hashmap-based datastructures to bypass dense GPU matrix multiplications, enabling complex model training on standard CPUs and drastically reducing compute costs.• Developed an In-Memory Inference Pipeline that dynamically routes queries to custom fine-tuned models based on document context, optimizing accuracy for domain-specific tasks.• Architected an end-to-end Retrieval-Augmented Generation (RAG) pipeline to automate document analysis, integrating OCR engines to extract structured data with 95% accuracy.• Built real-time data streaming pipelines using Apache Kafka and Apache Flink, processing over 1TB of data daily for anomaly detection microservices.

Education

• Chandigarh Group of Colleges, Landran, Punjab

  Bachelor of Technology - BTech, Computer Science

  2020 — 2024