Hello! 👋

I'm Sushant Gautam

AI/ML Engineer | Agentic Systems & MLOps

About Me

I am an AI/ML Engineer who owns the entire model lifecycle—from translating business ambiguity into concrete mathematical problems to deploying scalable, low-latency systems in production. My approach combines rigorous research (MS in CS, 4.0 GPA) with production discipline (AWS, CI/CD). Whether architecting 'System 2' agentic workflows or optimizing inference costs for serverless deployments, I focus on building robust, observable AI products that solve real problems at scale.

Agentic AI LLMs RAG LangChain Model Context Protocol (MCP) Generative AI PyTorch Computer Vision Deep Learning Diffusion Models GANs Scikit-learn TensorFlow MLOps LLMOps Docker Kubernetes MLflow DVC Airflow CI/CD AWS SageMaker AWS CDK GCP BASH Terraform Python FastAPI SQL TypeScript Next.js C++ CUDA Data Engineering Kafka PySpark Git Vector Databases Prometheus Grafana

Projects

01

🧩 Cognitive-Reflex-Agent

A 'System 2' reasoning agent capable of writing and executing Python code in a secure sandbox to solve complex queries. Features a semantic caching layer ('System 1') to recall past executions, significantly reducing latency and inference costs.

Agentic AIModel Context Protocol (MCP)Code InterpreterSecurity SandboxingSemantic CachingGemini 2.0
02

🧠 Context-Aware RAG Memory

A persistent memory system that 'learns' from user browsing. Uses hybrid temporal-semantic retrieval to recall exact information snippets, integrated with a Chrome extension for real-time augmentation.

RAGVector DatabasesFastAPIChrome ExtensionsEmbeddingsOllama
03

📈🤖 AI Stock Advisor

Multi-agent financial analyst platform. Orchestrates specialized LLM agents to fetch real-time Yahoo Finance data, analyze market trends, and generate personalized investment reports.

Multi-Agent SystemsPythonStreamlitFinancial Data EngineeringPrompt Engineering
04

🧩⚡ TabGist — Chrome Summarizer

Production-ready Chrome Extension for summarizing web content and YouTube transcripts. Built with a local-first philosophy using Gemini Flash for sub-second latency.

Chrome MV3JavaScriptFastAPINLPGenAIREST APIs
05

🩺 Medical RAG Chatbot

HIPAA-conscious RAG system for medical Q&A. retrieval accuracy optimized using LangChain and Pinecone, deployed via AWS CI/CD pipelines for reliable 24/7 availability.

RAGLangChainPineconeAWSFlaskHealthcare AI
06

End-to-End MLOps Pipeline

Full-cycle MLOps implementation for image classification. Automates data versioning (DVC), experiment tracking (MLflow), and model deployment to AWS SageMaker endpoints.

MLOpsAWS SageMakerMLflowDVCGitHub ActionsPyTorch
07

Real-Time Data Streaming Architecture

High-throughput data engineering pipeline processing live streams. Orchestrates Kafka producers/consumers and PySpark transformations within Dockerized Airflow environments.

Data EngineeringApache KafkaAirflowPySparkCassandraDocker
08

Self-Driving Car (End-to-End DL)

Autonomous steering system using Convolutional Neural Networks (CNNs). Clones human driving behavior by predicting steering angles directly from raw camera pixels.

Deep LearningComputer VisionOpenCVPyTorchAutonomous Vehicles
09

Handwritten Text Recognition (OCR)

Custom CRNN (CNN + LSTM + CTC Loss) architecture for reading handwriting. Achieved state-of-the-art accuracy on the IAM dataset using advanced data augmentation techniques.

TensorFlowOCRSequence ModelingComputer VisionData Augmentation
010

ROS2 Autonomous Tractor

Navigation stack for agricultural robotics. Implements sensor fusion (LIDAR + GPS) and path planning algorithms using ROS2 for precise field traversal.

ROS2RoboticsSensor FusionLocalizationPath Planning

Experience

SDE Intern (Transcribe Team)

Amazon Web Services (AWS)

May 2025 – Jul 2025
  • Architected a Next.js/TypeScript observability dashboard to visualize 1,000s of integration tests, reducing debug time by 50%.
  • Engineered a zero-touch CI/CD pipeline using AWS CDK & Elastic Beanstalk, slashing deployment time from hours to <10 minutes.
  • Developed AI-driven failure summarization tools, accelerating root-cause analysis by 40% for the engineering team.

Graduate Research Assistant

Mississippi State University

Aug 2023 – May 2025
  • Collaborated with researchers from **Stanford & Princeton** to develop diffusion models for image restoration, improving quality by 20%.
  • Built reproducible ML training pipelines using Docker, Kubernetes, and PyTorch-Lightning, optimizing hyperparameter tuning with Optuna.

Machine Learning Engineer

Proxmed Pty. Ltd.

Feb 2023 – Aug 2023
  • Deployed LVO detection models (YOLOv7) on AWS Lambda (Serverless), achieving 8% higher accuracy than legacy systems.
  • Optimized model inference using ONNX quantization and automated deployments via GitLab CI/CD.

ML Engineer & Researcher

Wiseyak / HeHealth

Oct 2020 – Apr 2022
  • Developed medical imaging diagnosis models, leveraging synthetic data generation (GANs) to overcome data scarcity.
  • Reduced false positive rates in skin lesion detection through few-shot learning techniques.

Education

M.S. in Computer Science (AI), GPA 4.0

Mississippi State University

Aug 2023 - Aug 2025
  • **Thesis:** X-DECODE — EXtreme Deblurring with Curriculum Optimization (arXiv, 2025).
  • **Leadership:** President, CSE AI Club (Organized GenAI workshops & hackathons).
  • **Coursework:** Cloud Computing, AI Robotics, Deep Learning, Advanced Algorithms.