Building Production-Grade AI Systems That Solve Real Problems
Samir Sengupta is an AI/ML Engineer passionate about building intelligent systems that solve real-world problems using Large Language Models, RAG architectures, and production-grade MLOps. He is developing SyGenticAI, a startup focused on building agentic AI systems where AI creates micro-agents capable of assembling entire software products.
His work spans designing and deploying production-ready LLM applications using RAG, embeddings, and semantic search; building scalable ML pipelines with distributed training and model optimization; and implementing cloud-native AI solutions on AWS Bedrock, Azure OpenAI, and GCP Vertex AI. Recent achievements include improving LLM reasoning accuracy by 35% on GSM8K and HumanEval benchmarks, reducing ML inference costs by 50% through quantization, and building microservices handling 10M+ daily events with 99.9% uptime.
Featured projects include PrometheusAI (privacy-first offline LLM assistant), NeuralScale (distributed training framework), and OmniGen (multi-modal generation engine). Samir brings a builder's perspective on what it takes to move AI from prototype to production — with systems engineered for reliability, scale, and real business impact.
Expertise
- •Large Language Models & RAG Architectures
- •Production MLOps & Cloud-Native AI
- •Agentic AI Systems
- •Model Optimization (Quantization, LoRA/QLoRA)
- •Distributed ML Training
- •Vector Databases & Semantic Search
Signature Moves
- •Founder — SyGenticAI (Agentic AI Startup)
- •35% Improvement in LLM Reasoning on GSM8K & HumanEval
- •50% Reduction in ML Inference Costs via Quantization
- •Built Microservices Handling 10M+ Daily Events
- •Projects: PrometheusAI, NeuralScale, OmniGen
- •AWS Bedrock, Azure OpenAI & GCP Vertex AI
“The future of AI belongs to engineers who can move from prototype to production — with systems that actually work at scale.”
– Samir Sengupta
