I build ML systems & LLM-powered agents, from production pipelines to agentic architectures. Master's in CS. AIE at Uber. Founder of RackX - an agentic job search assistant, and Cnotes (50K+ users) - A Business CRM for LCO's housing more than 50k+ users.
Developed predictive models to identify high-engagement participants and improve retention; built a RAG system over program data (curricula, grants, jobs, records) for natural language querying; and fine-tuned LLMs on noisy real-world data to improve robustness and domain alignment.
Machine Learning Engineer
Dell Technologies
2022 – 2023
ML pipelines for event data processing (XGBoost, scikit-learn). Real-time inference APIs in FastAPI, sub-200ms latency. Precision/recall evaluation with A/B-style comparisons.
Project Engineer - Full Stack & Cloud
Wipro
2021 – 2022
Java backend services on AWS. 40% API latency reduction. Docker + Kubernetes for service reliability at scale.
Junior Data Scientist
Robosoft Technologies
2020 – 2021
At Robosoft, I built and deployed machine learning solutions for mobile app analytics, from large-scale data pipelines and EDA to predictive models that improved system performance, automated workflows, and reduced operational bottlenecks.
Founder & Lead Engineer
Cnotes.in
2019 – 2021
Solo-built a full-stack CRM (LAMP + React) for Local Cable Operators — scaled to 50K+ users. Idempotent billing, audit-ready event logs, role-based auth.
Research & Publications
i.
In Progress · arXiv Extension
Stability-Ranked Sequential Merging for Agentic Planning Graphs
Extends TAPE's batch graph construction with an anchor-scored sequential merge pipeline. Novel edge-retention policy and ILP feasibility formulation for constrained LLM agent planning under tool-guided adaptive execution.
LLM AgentsPlanningILPGraph Merging
ii.
IEEE · Published
Prediction of CERN Electron Mass Collision using CATBoosting and LGBMR
Ensemble ML applied at particle physics scale. CatBoost + LightGBM with advanced feature engineering — 99% accuracy on CERN electron mass collision prediction.
CatBoostLightGBMEnsemble MLPhysics
iii.
IJNRD · Published
Detecting Diabetic Retinopathy using Deep Learning
Fully convolutional neural network for early-stage retinopathy screening. 93% accuracy, end-to-end clinical usability with TensorFlow and Django.
FCNTensorFlowMedical AI
iv.
IEEE · Published
Cirrhosis disease classification by using polynomial feature and XGBoosting
Cirrhosis disease classification using polynomial feature engineering and XGBoost, improving diagnostic prediction accuracy. Applied machine learning techniques to enable early detection and cost-effective diagnosis in healthcare settings.
Tejas Belakavadi Kemparaju
AI/ML Engineer · Researcher
