AI/ML Engineer & Data Scientist (5+ yrs)

SAI BHUVANESH SURYADEVARA

404-***-**** ********************@*****.*** linkedin.com/in/suryadevarasai/ Portfolio Atlanta, GA(Open to Relocate) SUMMARY

• AI/ML Engineer & Data Scientist with 5+ years of experience architecting production-grade systems across Sports-Tech, Banking, and Healthcare.

• Expert in building high-impact Computer Vision pipelines (YOLOv8, Pose Estimation) and Agentic LLM/RAG workflows (LangGraph, LlamaIndex) that drive measurable revenue.

• Proven track record of quantifying $1.5M+ in TV Value for sponsors and reducing manual workflows by 85% through scalable automation on AWS and Databricks.

EDUCATION

Master’s in Computer Science – Georgia State University, Atlanta, USA. Aug 2023 – Dec 2025 (GPA: 3.9/4.0) Bachelor’s in Computer Science – SRM University, India Jun 2019 - Jul 2023 (GPA: 3.8/4.0) Certifications: Databricks Generative AI Engineer, Azure AI Fundamentals, Introduction to Lang Chain & LangGraph TECHNICAL SKILLS

• Languages: Python, SQL, R

• ML/CV Libraries: PyTorch, TensorFlow, Keras, OpenCV, Scikit-learn, YOLOv8, ByteTrack, FAISS, NumPy, Pandas, Matplotlib

• AI/LLM Stack: LangChain, LangGraph, Llamalndex, Crew Al, RAG, LLM Fine-tuning (QLORA), Prompt Engineering

• Data Engineering: Databricks, Delta Lake (Medallion Architecture), Apache Spark, MLflow, Snowflake, Oracle DB, MongoDB

• Cloud & MLOps: AWS (S3, EC2, SageMaker, QuickSight), GCP (Vertex Al), Docker, Kubernetes, CI/CD, Git, Jenkins

• Other Tools& APIs: Power BI, Tableau, REST APIs, Microservices, FastAPI EXPERIENCE

SportsBiz – Atlanta, USA Jan 2026 – Present

Data Scientist/CV Engineer (Sports-Tech)

• Designed a multi-agent workflow (LangGraph + Llamalndex) that combines CV-derived logo exposure metrics with broadcast metadata to assemble sponsor ROI reports automatically — reduced report turnaround from 2-3 days of manual analysis to under an hour. Reports feed directly into client briefings.

• Built a real-time brand exposure pipeline using YOLOv8 + pose estimation that processes live broadcasts frame-by-frame - automated analysis of 227k+ impressions with ByteTrack and developed an internal scoring system across 9 visual quality dimensions (size, clarity, screen clutter,)

• Optimized a three-stage inference pipeline on Databricks using a Medallion Architecture with Delta Lake; added a custom Camera Change Detection module to handle broadcast cuts cleanly — brought unique player and brand impression accuracy to 98%.

• Projected pose-tracking coordinates onto 3D stadium models in AWS to spatially map brand exposure by camera angle and field zone— analysis contributed to quantifying $1.5M+ in total TV Value across campaigns.

• Built the LIP Model — a player identification ensemble combining OCR, face recognition, and dynamically updated event rosters; cut manual labeling time by 85%, making end-to-end processing of full broadcast events practical at scale. Georgia State University – Atlanta, USA Aug 2024 – Dec 2025 Research Data Engineer /Data Scientist

• Built a RAG-based academic advising assistant using LangChain, OpenAl embeddings and FAISS — grounded responses in live institutional policies and course materials, measurably reducing out-of-context answers on an internal eval set.

• Built a document ingestion pipeline with OCR, semantic chunking, metadata enrichment, and vector indexing to digitize legacy academic records — expanded the system's searchable document base by 40%.

• Fine-tuned a multimodal LLM using PyTorch for structured academic report generation from transcript and scanned document inputs; achieved a 12% improvement in coherence scores on internal benchmarks with inference latency under 2 seconds. Vensai Technologies – Atlanta, USA Jun 2024 – Aug 2024 Data Scientist (Banking)

• Developed an automated vendor contract auditor using LangGraph and Milvus 2.4 to extract SLA and liability clauses from logistics contracts — achieved 98% extraction accuracy through Corrective RAG (CRAG) workflows.

• Built a document processing pipeline using Unstructured.io and LayoutLMv3 to digitize legacy freight invoices and bills of lading - expanded searchable institutional document coverage by 35% through high-fidelity metadata extraction.

• Fine-tuned a Llama 3 model via QLoRA for high-risk legal clause detection (+12% F1); deployed on Kubernetes with TensorRT-LLM achieving sub-2-second inference in production.

Georgia State University – Atlanta, USA Aug 2023 – May 2024 Research Data Engineer /Data Scientist (Banking)

• Collaborated with a banking enterprise to build loan default risk models (XGBoost, Random Forest, PyTorch) hosted on AWS SageMaker and scaled via Kubernetes — improved risk detection by 12% while cutting cloud infrastructure costs by 18% through efficient training configuration.

• Conducted statistical analysis and bias detection across multi-source banking datasets (credit, customer, transaction) to engineer a feature library used across all downstream models.

• Built AWS QuickSight dashboards to visualize loan disbursement, delinquency, and portfolio risk metrics - gave the team a self-serve view of portfolio health that previously required ad-hoc SQL reports. HCL Technologies – Vijayawada, India Jan 2021 – Jul 2023 Jr. Data Scientist (Healthcare/Insurance)

• Developed predictive models for healthcare utilization using LSTM (TensorFlow) on EHR data and XGBoost on claims data, integrating engineered features to achieve an 8% reduction in hospital readmission rates through engineered clinical features.

• Managed large-scale ETL workflows using Snowflake, SQL, and Python to prepare and transform multi-source data for use in ML pipelines and executive dashboards, improving data quality and model readiness.

• Optimized SQL query execution plans for large log tables - reduced query runtime by 30% and accelerated the research iteration cycle.

• Built Power BI dashboards with Power Query and DAX to monitor healthcare utilization and cost trends; automated CI/CD for microservices using Jenkins and Docker, eliminating manual deployment steps. ACHEIVMENTS

• Received Gold Medal and Special Mention Award for Research at SRM University, India

• Secured 2nd place in a 24-hour national ML hackathon at IIIT Sri City, India.

• Nominated for the Chancellor's award at SRM for excellence in academics and outstanding contribution to the university.

• Received ‘Star Performer’ award for my work on ML-driven fraud detection model.

• 100% scholarship awardee for academic merit and research in Masters.

Contact this candidate