Available immediately for full-time AI/SWE roles

Ishan Shirode Production AI Engineer

Bridging ML research with robust backend engineering. I architect and deploy scalable AI systems—from high-speed RAG pipelines to automated MLOps architectures.

[Add profile.png to public/]Ishan Shirode
5+AI Systems Built
2Industry Experiences
4Production Projects
2026Graduate

About Me

I specialize in building production-grade AI systems using RAG, FastAPI, MLOps, and LLM orchestration.My focus is moving AI solutions beyond notebooks and into scalable, deployable systems that drive real business value.

/bento_grid_ml_pipeline.png
AI Systems Engineer
Engineering ML pipelines from data ingestion to containerized deployment.
Global Engineering Mindset
Available for full-time AI/SWE roles globally
Python, FastAPI, Docker, PyTorch, LangChain, FAISS
My Tech Stack
PythonLLMsFastAPI
TransformersSQLDocker
Infrastructure Focus
Productionizing ML pipelines with Docker, FastAPI, and Airflow.
Current Technical Focus
Orchestrating multi-agent LLM consensus and FAISS vector retrieval.
Engineering Philosophy
Bridging ML research with highly scalable backend engineering.

My Skills

Artificial Intelligence

Machine Learning
Natural Language Processing (NLP)
Large Language Models (LLMs)
Generative AI
Transformers
RAG Architectures
Prompt Engineering

Expertise & Tools

Computer Vision (YOLOv8)
OpenCV
OpenAI Whisper
HuggingFace
LangChain
TensorFlow
Scikit-Learn

MLOps & Engineering

FastAPI
Docker
AWS Cloud (EC2/S3)
MLOps Pipelines
GitHub Actions
Linux
Postman

Data & Programming

Python
C++
SQL
Pandas
NumPy
Matplotlib
Seaborn
Business Analytics

Professional Skills

Problem Solving
System Design
Technical Leadership
Agile Methodology

Engineering Activity

120+

Contributions

15

Repositories

Currently Building

Multi-LLM Council

High-Impact AI Solutions

AI-Powered Portfolio Platform

AI-Powered Portfolio Platform

Latency<1s
Vector DBFAISS

A full-stack RAG and Agentic LLM platform that transforms raw developer data into a recruiter-ready experience.

Next.js
FastAPI
Gemini
FAISS
Read Case Study→
AI Knowledge Assistant (RAG Engine)

AI Knowledge Assistant (RAG Engine)

Latency<1s
Vector DBFAISS

A production-grade Retrieval-Augmented Generation engine executing hybrid search across academic corpuses.

FastAPI
FAISS
LangChain
Docker
Read Case Study→
DriftGuard-ML

DriftGuard-ML

Drift Reduced40%
Reliability99.9%

A closed-loop MLOps pipeline featuring automated drift detection, alerting, and Champion-Challenger retraining.

Celery
Prometheus
FastAPI
Read Case Study→

How I Build AI Systems

1. Enterprise RAG Pipeline

Documents
→
Semantic Chunker
Embedding Model
Uses BGE-large-en-v1.5 for dense representations, scoring top marks on MTEB benchmarks.
→
FAISS Vector DB
Chose FAISS for its optimized C++ backend, enabling <50ms nearest-neighbor search on 1M+ vectors.
User Query
→
Retriever
↓
LLM Generation
→
Final Output

2. Automated MLOps (DriftGuard)

Live Ingestion
→
FastAPI Inference
↓
Prometheus Metrics
Time-series DB tracking confidence scores, latency, and input distributions.
→
Drift Detector
Calculates KS-Statistic on incoming feature batches. Triggers if p-value < 0.05.
↓
Alert Trigger
→
Airflow Retraining

3. Multi-LLM Council

User Prompt
→
LangChain Router
Uses semantic routing to determine if the query requires standard retrieval, or multi-model consensus.
GPT-4
Claude
Llama-3
↓
Evaluator (Consensus)
A smaller, fast model (like Llama-3-8B) that acts purely as a judge to merge and verify the multi-node outputs.
→
Verified Answer

4. Scalable Inference API

Client Request
→
Load Balancer
↓
FastAPI Async Worker
↔
Redis Cache
↓
PyTorch Engine

My Education

Leadership & Activities

Anti-Ragging Head

Student Council

Joint Digital Head

AS&H Department