resume
Bhuvan Chennoju
experience
Senior Data Scientist — ML Systems, Fraud & RiskT-Mobile USA
2024-09 — present- ·Built three production fraud classifiers (first-party fraud, third-party equipment fraud, ATO) with score-based decision thresholds; deployed via MLflow on Databricks with Azure Data Factory orchestration.
- ·Designed a grounded LLM-as-judge agent (Azure OpenAI) to triage ML-flagged accounts in the human review queue, with MLflow tracing for evaluation and drift monitoring.
- ·Built an NLP + graph threat intelligence pipeline processing 1-2M messages daily: RAPIDS HDBSCAN clustering, topic modeling, LLM enrichment, and graph linkage across sender IDs, URLs, and cell-tower signals.
- ·Leading a MILP-based least-cost transaction routing system for a $3.35B/month, 20M+ transaction portfolio — backtesting shows $8M annual savings against a $16M target.
- ·Designed Unity Catalog feature store schemas and GitLab CI/CD structure for the ML team; contributed to a semantic chunking RAG system on Databricks Vector Search.
- ·Mentored 5 data scientists and analysts on modeling, code quality, and interpretability.
Data Scientist — ML Research & Data EngineeringStowers Institute for Medical Research
2023-08 — 2024-08- ·Pretrained a masked language model (GenLM) for genomic sequences from scratch — 12-layer dilated CNN on 13 Drosophila species genomes with a custom character-level tokenizer, multi-GPU sharded dataloader, and W&B experiment tracking.
- ·Built a parallelized DAG-based ETL pipeline processing 13 genomes (~2.5 GB) in under 1 hour; trained a multi-task CNN for DNA-protein binding site prediction, improving accuracy 15% over baseline.
Graduate Research Assistant — Representation Learning & BiometricsComputational Intelligence & BI Lab, UMKC
2021-10 — 2023-07- ·Developed deep learning models for privacy-preserving biometric identity verification using vascular-pattern imaging and representation learning.
- ·Co-inventor on U.S. Patent Application No. 24UMK017 (2024); published at IEEE IJCB 2023 and BIOSIG 2023.
Software Developer Intern — Machine LearningT-Mobile USA
2022-05 — 2022-08- ·Benchmarked deep learning classifiers for 5G edge vs. cloud inference (13% latency reduction on edge hardware); built gesture recognition and NLP voice-control endpoints for a Unity VR app on Oculus Android.
Data Scientist — Computational Fluid Dynamics & MLIndian Institute of Technology Bombay
2019-06 — 2021-08- ·Developed OpenFOAM CFD simulations on HPC clusters; trained an ML surrogate model to predict atomization droplet size distributions from nozzle geometry — replacing expensive simulation runs with fast inference.
- ·Published in Physics of Fluids (AIP, 2020) — 14 citations. DOI: 10.1063/5.0020518.
Data Analyst — Manufacturing & IoT AnalyticsLarsen & Toubro Ltd
2018-07 — 2019-06- ·Analyzed IoT sensor data from manufacturing facilities to identify downtime patterns and predict failure events; built real-time equipment health dashboards.
education
M.S., Computer Science (AI/ML)University of Missouri – Kansas City
- ·Thesis: Graph-based neural networks for time series forecasting.
B.Tech, EngineeringNational Institute of Technology Hamirpur
certifications
Microsoft Certified: Azure Data Scientist AssociateMicrosoft
2023 — presentMicrosoft Certified: Azure Data Engineer AssociateMicrosoft
2023 — presentresearch & publications
Deep Vascular Personal Identification SystemU.S. Patent Application No. 24UMK017 — UMKC
2024 — present- ·Co-inventor. Deep learning system for privacy-preserving biometric identity verification using vascular-pattern imaging.
Analysis of fNIRS as a Biometric ModalityIEEE International Joint Conference on Biometrics (IJCB)
2023 — present- ·Chennoju B., Bajaj K., Rahman M., Derakhshani R.
A Wrist-Worn Diffuse Optical Tomography Biometric SystemBIOSIG 2023 — IEEE/GI
2023 — present- ·Akula S. et al.
Novel Flame Dynamics in Rich Mixture of Premixed Propane–Air in a Planar MicrocombustorPhysics of Fluids 32(10) — AIP
2020 — present- ·Chennoju B. et al. 14 citations. DOI: 10.1063/5.0020518.