Sayan Ghosh

AI Researcher

India0 mo experience

AI EnabledAI ML Practitioner

Key Highlights

Expert in AI/ML with a focus on agentic systems.
Proven track record in developing scalable cloud solutions.
Strong background in conversational AI and NLP.

Stackforce AI infers this person is a skilled AI/ML developer with expertise in cloud architecture and conversational systems.

Contact

sayan18012004@gmail.com LinkedIn

Skills

Core Skills

Retrieval-augmented Generation (rag)Database InteractionAgentic AiLarge Language Models (llms)Conversational AiNatural Language Processing (nlp)Aws Cloud ArchitectureCloud Security Best Practices

Other Skills

Systems DesignSQLAzureAPIsLangChainLangGraphReinforcement LearningDockerLLMsGenAIAgile MethodologiesChatbot DevelopmentAWSCloud ArchitectureInfrastructure Security

About

Always eager to connect, collaborate, and innovate. Let’s build something impactful!

Experience

0 mo

Total Experience

Average Tenure

Current Experience

Sereno

AI/ML Developer

Aug 2025 – Sep 2025 · 1 mo

● Worked closely with clients to build a multi-agent RAG system that integrates SQL data with Azure workflows for large-scale, collaborative information retrieval.
● Refactored backend pipelines by replacing fuzzy matching with deterministic orchestration via Microsoft’s Semantic Kernel, significantly improving SQL query accuracy and eliminating frequent table mis-selections.
● Built an automated ingestion pipeline for Azure Blob Storage data with file-agnostic row-level deduplication, removing all manual indexing, saving 30+ engineering hours/month, and enabling scheduled Azure AI Search updates.
● Integrated modular APIs across multiple backends, reducing deployment friction and ensuring seamless cross-project interoperability.
● Authored schema documentation for 13+ database tables, accelerating onboarding and improving retrieval precision in LLM-driven SQL query generation.

Retrieval-Augmented Generation (RAG)Database InteractionConversational AISystems DesignAgentic AISQL+2

Innotrat labs

AI ML Developer

May 2025 – Jun 2025 · 1 mo

● Worked at a deep-tech lab focused on agentic GenAI, delivering enterprise-grade local AI systems without reliance on cloud APIs.
● Led end-to-end development of a modular, role-based GenAI platform using LangChain + LangGraph and state-driven LLM orchestration.
● Implemented custom RL loops with Unsloth (LoRA + paged attention), enabling on-device fine-tuning on a 16GB GPU.
● Achieved 100% local hosting by deploying LLMs through Ollama and LM Studio, removing all external API dependencies.
● Dockerized the entire stack, integrated secure auth, and helped engineer the system for horizontal scalability to support >100,000 users.
● Result: Delivered a plug-and-play, open-source ready framework for running agentic LLM systems on air-gapped hardware.

LangChainLangGraphReinforcement LearningDockerLLMsGenAI+2

Infosys springboard

AI Intern

Oct 2024 – Dec 2024 · 2 mos

● Contributed to the Infosys R&D team by building Meditrain, a client-facing GenAI medical chatbot under real-world constraints.
● Spearheaded the design of a RAG-based chatbot with context memory and multilingual NLP pipelines.
● Improved classification accuracy by 30%, reduced hallucinations by 20%, and boosted user engagement by 25%.
● Integrated secure APIs for accessing verified medical data, building safeguards to meet compliance and privacy needs.
● Led weekly Agile sprints, shortening deployment cycles by 40% and accelerating the QA-feedback loop.
● Proposed roadmap features like dynamic memory optimization and scaling to non-English queries (Hindi, Bengali, German, Spanish).

Conversational AINatural Language Processing (NLP)Agile MethodologiesChatbot Development

Aicte-eduskills & aws academy

AWS Cloud Virtual Intern

Oct 2024 – Dec 2024 · 2 mos

● Finished a Government-sponsored internship in collaboration with AWS Academy, focused on architecting and securing scalable cloud infrastructure in enterprise-like environments.
● Designed and deployed high-availability, auto-scaling cloud infrastructure using EC2, S3, RDS, VPC, and Load Balancers.
● Reduced theoretical cloud billing by up to 20% through efficient resource provisioning and scaling policies.
● Hardened infrastructure with IAM-based access controls, encryption at rest, and isolated VPC subnets.
● Delivered cloud blueprints that match enterprise deployment patterns for microservice-based apps.

AWSCloud ArchitectureInfrastructure SecurityAWS Cloud ArchitectureCloud Security Best Practices