Loading Portfolio
Ayush Deshmukh · IISc Research Intern · AWS SA Certified
CS undergrad at VIT & AI Research Intern at IISc Bangalore — building production LLM microservices, RAG pipelines, and DevSecOps automation. AWS Certified Solutions Architect.
About
CS undergraduate at VIT Bhopal and AI Research Intern at IISc Bangalore. I specialise in Generative AI engineering, scalable backend systems, and LLM-powered agentic pipelines — with a strong focus on reliability, observability, and shipping software that stays robust under real use. AWS Certified Solutions Architect.
Intro
B.Tech Computer Science at VIT Bhopal and active AI Research Intern at IISc Bangalore working on LLM microservices and agentic RAG systems. AWS Certified Solutions Architect – Associate. My interests centre on robust software, backend quality, CI/CD, cloud-native development, and AI-enabled developer tooling.
2023
Started B.Tech at VIT
2027
Expected graduation
Core Stack
Community
I enjoy sharing what I learn through community events, developer circles, and hands-on collaboration. Open source contribution and Google Developer Student Clubs work have shaped how I think about mentorship, documentation, and shipping in public. As a Technical Member in Google Developer Students Club (2024-2025), I participated in peer learning sessions and collaborative problem-solving activities.
Working Style
Current Focus
Technical Direction
My stack spans Python, Java, C++, SQL, FastAPI, LangChain/LangGraph, vector databases, AWS, Docker, and CI/CD automation. I build systems that are observable, testable, and easy to operate in production.
LangGraph · RAG · Qdrant · Gemini API
LangChain, LangGraph, Retrieval-Augmented Generation, FAISS, Qdrant, LLM Inference, Prompt Engineering, Agentic AI Systems, and multi-step reasoning pipelines for production research environments.
FastAPI · AWS · Docker · PostgreSQL
FastAPI, REST API Design, PostgreSQL, MongoDB. AWS (EC2, Lambda, S3), Docker, CI/CD Pipelines, GitHub Actions. Languages: Python, Java, C++, SQL — with a focus on reliability and maintainable service design.
Pytest · GitHub Actions · Bandit · CI/CD
Automated testing, Pytest, GitHub Actions, Bandit SAST, structured logging, debugging, and deployment workflows. 90%+ test coverage on research codebases with zero-downtime deployments.
Certifications & Achievements
Featured Projects
Each project enters one-by-one as you scroll, with more room for the systems thinking, implementation choices, and outcomes behind the build.
Production-grade retrieval-augmented generation
Full-stack RAG architecture improving document recall by 38% via hybrid BM25 + Qdrant dense vector retrieval. Cross-encoder reranking (MS-MARCO MiniLM) boosts top-5 precision to 91%. Sub-800ms end-to-end LLM response via a containerised FastAPI backend on Cerebras hardware.
AI-driven DevSecOps test-fix pipeline
AI-driven DevSecOps pipeline with 85% first-pass code generation success via autonomous test-fix loops using LangGraph. Redis semantic caching cut LLM costs 40% and latency from 10s → 50ms. Bandit SAST integration flags 15+ vulnerability classes across 200+ automated test runs.
Fault-tolerant API-driven workflow backend
Fault-tolerant backend reducing unhandled failures by 60% and mean recovery time by 45% via configurable retry logic and structured logging. 12 RESTful API endpoints over a normalised PostgreSQL schema (8 tables). Docker Compose containerisation cut local setup time by 70%.
Self-correcting explainer video generator
Modular AI pipeline turning concepts into rendered animated explainer videos using Manim, with automated visual QA via OpenCV and synced voiceovers. Self-healing render loop validates each frame before assembly, eliminating manual review cycles.
-- PROFESSIONAL EXPERIENCE
APR 2026 – PRESENT
ActiveIndian Institute of Science (IISc), Bengaluru
CINT LAB (AE129) · DR. S. N. OMKAR
Social Proof
The reliability work and product thinking were equally strong. The result felt both engineered and designed.
Riya Patel
CTO, FinEdge
Fast iterations, strong systems intuition, and a frontend eye that made the entire experience feel premium.
Miguel Anders
VP Engineering, AiRobotics
A rare mix of technical depth and presentation craft. Complex workflows suddenly felt understandable.
Lea Norton
Product Head, OmniWork
Initiate Contact
Currently open for new opportunities. Whether you have a question or just want to say hi, I'll try my best to get back to you!
ayushdeshmukh.dev@gmail.com
Github
Ayush1Deshmukh
Ayush Deshmukh
Location
Bhopal, Madhya Pradesh, India