Available for opportunities
Deep Learning Engineer specializing in GPU Optimization and LLM Inference. MS in Computer Science from UT Arlington (May 2025).
I optimize ML systems for speed, memory, and cost. My work focuses on CUDA kernels, FlashAttention, quantization techniques, and building production-grade inference infrastructure.
Published researcher with work accepted at IEEE ICC 2026 on LLM-enabled path planning for wireless networks.
Python
PyTorch
TensorFlow
CUDA
C++
Java
Go
Hugging Face
Docker
Kubernetes
AWS
PostgreSQLPythonPyTorchTensorFlowCUDAC++JavaGoHugging FaceDockerKubernetesAWSPostgreSQL
Redis
MongoDB
Airflow
Spark
Linux
Git
Neo4j
FastAPI
Flask
NumPyRedisMongoDBAirflowSparkLinuxGitNeo4jFastAPIFlaskNumPyAI Solutions Engineer Intern
Mar 2026 - May 2026 โข New York, NY (Remote)
Configured LLMs for protocol-specific clinical workflows, orchestrating radiologist report parsing and EMR data extraction across EPIC/FHIR-integrated hospital systems including Medstar, Mount Sinai, and UFL. Deployed and maintained AI inference endpoints supporting 6+ live health system sites under real-time US time zone SLAs. Prepared HIPAA-aligned technical documentation including DPIAs, security questionnaires, and EPIC bidirectional integration guides for clinical AI deployment.
Machine Learning Engineer Intern
Dec 2025 - Feb 2026 โข United States
Built computer vision pipelines for intra-oral image analysis across 6+ clinical categories (gingivitis staging, plaque detection, recession classification) on a 50K+ labeled dataset. Engineered automated retraining pipeline on AWS SageMaker incorporating dentist-corrected labels, improving model accuracy iteratively across production inference endpoints.
Graduate Research Assistant
Jun 2025 - Present โข Arlington, TX
Built TopGPT, a full-stack LLM application fine-tuned on 3+ textbooks with RAG over 1,000+ research paper chunks stored in Pinecone on AWS. Also contributed to CTMap, an LLM-enabled path planning system for mmWave 6G networks (accepted at IEEE ICC 2026), fine-tuning LLMs on Dijkstra-generated paths with real-time OpenStreetMap + Sionna 6G integration.
Graduate Teaching Assistant
Aug 2024 - May 2025 โข Arlington, TX
Supported graduate courses in Numerical Methods for 50+ students, assisting with algorithmic problem solving, optimization, and computational modeling. Concurrently developed CTMap, an LLM-enabled 6G path planning system accepted at IEEE ICC 2026, fine-tuning LLMs on Dijkstra-generated coordinate paths and applying them to Sionna wireless network simulation outputs.
Senior Software Engineer โ Software Engineer
Jun 2019 - May 2023 โข 4 years โข Chennai
Designed and owned Java-based distributed data processing services handling millions of records daily across production systems serving 10+ enterprise clients. Led system design and architecture reviews for data-intensive microservices. Built fault-tolerant backend pipelines with high availability and reduced processing latency by 40% through service refactoring.
Master of Science - Computer Science & Engineering
๐ Specialization in Deep Learning
Aug 2023 - May 2025
Completed 10+ rigorous graduate-level courses, building a solid foundation in advanced computing and ML systems. Graduated May 2025 with 4.0 GPA.
Graduate Coursework
Bachelor of Technology - Computer Science Engineering
2015 - 2019
CTMap: LLM-Enabled Connectivity-Aware Path Planning for mmWave 6G Networks
Fine-tuned LLMs on Dijkstra-generated coordinate paths from OpenStreetMap, applied to Sionna 6G wireless simulation outputs to produce real-time optimal paths for mmWave connectivity-aware routing.
Two production-grade systems: (1) Custom CUDA C++ scaled-dot-product attention kernel with tiled QKแต, numerically stable softmax, and pybind11 PyTorch binding โ 5.64x faster than PyTorch at N=32, correctness verified at max_diff < 1e-7. (2) GPT-2 ported to AWS Inferentia via neuronx-cc 3-step pipeline (TorchScript trace, neuronx-cc compile, profiling) achieving 3.7x speedup (45ms to 12ms, 1,800 to 6,700 tokens/sec). Deployed as FastAPI REST endpoint on HuggingFace Spaces.
Supervised fine-tuning of Qwen2.5-7B using LoRA (r=8, alpha=16) on UltraFeedback, training only 0.5% of parameters (35M of 7B) with QLoRA 4-bit quantization and FP16 mixed precision. Achieved 17% training loss reduction (1.412 to 1.176) and 0.855 BERTScore in 30 minutes on a T4 GPU. Model merged and deployed to HuggingFace Hub.
Benchmarking framework comparing 4 attention implementations on NVIDIA L4 (seq len 1024, batch 32): FlashAttention-2 achieves 12.3x throughput (573K to 6.03M tok/s) and 99.7% memory reduction (12,582 MB to 38 MB) vs vanilla. Includes batch-size auto-tuner using binary search and ONNX/TensorRT export benchmarks. Key finding: algorithm-level optimization (FlashAttention) outperforms hardware-level optimization (TensorRT) by 6x for attention operations.
Multi-strategy reasoning system implementing 4 research papers: Chain-of-Thought with self-consistency voting (3 independent paths), Tree-of-Thoughts with beam search (width=3, depth=3), ReAct with real-time web search via Tavily API and ChromaDB vector memory, and Multi-Agent collaboration (Planner, Worker, Critic). LLM-based auto-classifier routes each query to the optimal strategy. Rate-limited (10/min, 100/day), with real-time streaming. Built with Groq LLM, deployed on HuggingFace Spaces.
UC Berkeley EECS โข Jul 2025
Amazon Web Services โข Dec 2024 - Dec 2027
Microsoft โข Aug 2025
Oracle Cloud โข Jun 2024 - Jun 2026
GPU optimization, CUDA kernels, FlashAttention, quantization techniques.
View ProjectsLLM fine-tuning, RAG pipelines, multi-agent systems, vLLM inference.
View ProjectsProduction ML systems, MLOps, model serving, Airflow pipelines.
View ProjectsDistributed systems, Java services, high-throughput APIs, databases.
View ProjectsAWS infrastructure, data pipelines, Spark, Kafka, ETL workflows.
View ProjectsIEEE publications, LLMs for wireless networks, digital twin simulations.
View Publications
