AI systems & agents
Tool-calling agents, MCP servers, and document workflows with structured interfaces, retries, and practical orchestration.
Software Engineer & AI Systems Engineer
Hi, I'm
Sayantan.
I build intelligent systems that work beyond the demo.
Let's talkTools of the tradeAlways learning. Always building.
Python ✳FastAPI ✳Flask ✳Java ✳Spring Boot ✳PostgreSQL ✳pgvector ✳Redis ✳Docker ✳MCP ✳LangGraph ✳AutoGen ✳PyTorch ✳WebSockets ✳Nginx ✳GitHub Actions ✳
GitHub Actions ↗Nginx ↗WebSockets ↗PyTorch ↗AutoGen ↗LangGraph ↗MCP ↗Docker ↗Redis ↗pgvector ↗PostgreSQL ↗Spring Boot ↗Java ↗Flask ↗FastAPI ↗Python ↗
01 / A little contextEngineer. Builder. Problem solver.
About me
I work at the intersection of AI systems and backend engineering. From retrieval pipelines and tool-calling agents to real-time rendering infrastructure, I turn complex ideas into reliable software. My work spans Python, Java, production APIs, and the systems that connect them.
Build something together50%+
Lower repetitive database-call costs
RAG pipeline · résumé-reported45%+
Lower AI token costs
Hybrid search · résumé-reported300+
LeetCode problems solved
Problem solving · résumé-reported02 / What I bringFrom architecture to delivery
Expertise
Retrieval & search
RAG pipelines, vector databases, and hybrid search that combine semantic retrieval with keyword relevance.
Backend engineering
APIs and application services in Python and Java, with database integration, asynchronous jobs, and clear interfaces.
Real-time infrastructure
WebSockets, server-sent events, live log fan-out, and session-aware delivery for responsive applications.
Deployment & delivery
Containerized services, reverse proxies, automated pipelines, and maintainable environment-based configuration.
03 / Selected engineeringSystems with a purpose
Projects
01
Real-time systems
Cake Cloud Backend
The idea
Rendering infrastructure. Connected in real time.
Built with
An asynchronous backend that connects rendering services with frontend clients through REST, server-sent events, and WebSockets.
Architecture / overview
Render engine→FastAPI gateway→Live client
Conceptual architecture · not a product screenshot02
AI / ML
AI Post-Processing Pipeline
The idea
From rendered pixels to new possibilities.
Built with
A modular AI post-processing pipeline connected to a physically based C++ renderer and controlled through a PyQt5 desktop interface.
Architecture / overview
C++ renderer→AI processing→Enhanced output
Conceptual architecture · not a product screenshot03
AI / ML
Tool-Calling AI Agent
The idea
Tools that turn language into action.
Built with
An agent that plans tasks, invokes tools, and synthesizes responses through a local OpenAI-compatible LLM endpoint.
Architecture / overview
User request→Agent + tools→Structured result
Conceptual architecture · not a product screenshot04
Backend
Render Job Orchestration
The idea
Queue. Render. Deliver.
Built with
A Flask REST service that dispatches rendering jobs to a custom C++ ray-tracing engine through a threaded job queue.
Architecture / overview
Job queue→C++ ray tracer→SFTP delivery
Conceptual architecture · not a product screenshot04 / The journeyExperience that compounds
Experience
Dec 2024 – Present
AI Systems Engineer
Microfacet.io · Non-Profit
MCP tools, retrieval pipelines, AI agents, model integration, and production delivery.
Aug 2024 – Present
Associate Consultant
Siemens Digital Industries Software
Teamcenter customization, backend and database features, production support, and CI security.
Jun 2023 – Dec 2023
Software Engineer
Zet · Previously OneCode
Java and Spring Boot backend services, MySQL integration, and performance optimization.
Feb 2023 – Jun 2023
Software Engineer Intern
Zet · Previously OneCode
Node.js webhooks, API reliability, and real-time service integrations.
Education
Vellore Institute of Technology
B.Tech · Computer Science Engineering / 2019 – 2023 / GPA 8.29