Sai Karthik
Indukuri
MS Artificial Intelligence · University at Buffalo
Deep Learning & Generative AI researcher building scalable, production-grade ML systems. Published NLP research, 14,000-student deployment, GPA 3.91.
Building AI that works
in the real world
I'm a graduate student at University at Buffalo pursuing an MS in Artificial Intelligence (GPA 3.91, expected May 2027). My work sits at the intersection of generative AI, scalable data systems, and applied ML — I care deeply about models that ship, not just models that score well on benchmarks.
Before grad school, I earned my BTech in Computer Science from NIT Andhra Pradesh and worked as an ML Engineer at Devtown, where I built an AI tutoring platform used by over 14,000 students. I'm also a published researcher — my NLP work was accepted at ICAAAIML-2025 and is under review at Springer.
Technical Expertise
Research & AI
- Generative AI (LLMs, GANs)
- NLP · Transformers · BERT
- Computer Vision (CNNs, ViTs)
- Retrieval-Augmented Generation
- Reinforcement Learning
- Causal Inference · Fine-tuning
Languages & Frameworks
- Python · C++ · Java · R · SQL
- PyTorch · TensorFlow
- JavaScript · React · Node.js
- MATLAB · scikit-learn
- FastAPI · Flask
Tools & Infrastructure
- AWS · Docker · Spark
- MLOps · CI/CD · CUDA
- Hugging Face · Databricks
- REST APIs · Git · LaTeX
Core Concepts
- Data Structures & Algorithms
- Convex Optimization
- Probability & Statistics
- Linear Algebra · Numerical Analysis
- Performance Engineering
Work History
Graduate Teaching Assistant
EAS587 — Data Intensive Computing · University at Buffalo
- Guided 50+ students in Hadoop and cloud architectures through grading and technical reviews.
- Evaluated Spark and MapReduce implementations, providing feedback on code scalability and runtime optimization.
Machine Learning Engineer
Devtown Pvt. Ltd. · Bengaluru, India
- Co-developed a real-time AI virtual tutor using Python, TensorFlow, and Flask — improving student learning outcomes by 16% as measured by quiz scores across the platform.
- Optimized the personalization algorithm via Pandas and SQL analysis, increasing AI-driven response relevance by 35%.
- Engineered a scalable production pipeline with scikit-learn and Docker on AWS EC2 GPU instances, serving 14,000+ active students.
Selected Work
Prose2Poetry
Engineered a novel NLP model to generate rhyming couplets from non-rhyming prose using FastText embeddings and a modified trigram Markov chain with phonetic analysis.
- 92% semantic similarity score
- 30% lower rhyme-pair error vs. bigram baseline
- Accepted at ICAAAIML-2025 · Under review at Springer
NYC Taxi Smart Agent
Distributed ETL pipeline processing 780,000+ NYC taxi records with NOAA weather integration, plus a natural language "Smart Agent" backend enabling non-technical users to query terabyte-scale data.
- R² of 0.66, RMSE of 10.92 on urban mobility prediction
- MCP-powered natural language query interface
- Identified distinct demand surges via "WeatherTax" meteorological features
CVE Threat Intelligence Pipeline
Production-grade Bronze-Silver-Gold Medallion Architecture on Databricks processing 40,000+ high-dimensional CVE records with full Delta Lake ACID compliance.
- 61,000+ affected products indexed via PySpark explode operations
- 100% data integrity via automated quality assertions
- Identified 38% surge in 2024 vulnerabilities
Academic Background
M.S. Artificial Intelligence
University at Buffalo, NY
B.Tech Computer Science
National Institute of Technology, Andhra Pradesh
Let's connect
Open to Summer 2026 internship opportunities in ML, AI research, and data engineering roles.