ML Systems & Backend Engineer · Building intelligent, data-intensive systems that scale
I design and ship production ML pipelines, real-time backend services, and scalable data infrastructure. From fine-tuning transformers (RoBERTa, ResNet-50) to deploying threat detection microservices processing 500K+ daily events — I build systems that deliver measurable impact.
📍 CS @ University of Virginia · Graduating May 2026 · Open to new-grad SWE roles
- Fine-tuned RoBERTa transformer for hate speech detection on 25K tweets — achieved 96.9% accuracy and 0.811 macro-F1, outperforming classical ML baselines
- Built a real-time threat detection microservice (Python/FastAPI) integrated into a Zero Trust security platform — reduced mean time to detect by 40%
- Engineered a data ingestion pipeline processing 500K+ daily OSINT events via Kafka + PostgreSQL with sub-second entity resolution
- Trained ResNet-50 CNN on 80K images with GPU-accelerated pipeline and Grad-CAM validation — 93% accuracy, 18% recall improvement
- Designed a geohash-based proximity matching system with dynamic precision, achieving sub-second queries while minimizing Firestore read costs
Proximity-based social matching platform — Co-Founder & Lead Engineer
- Designed a geohash spatial indexing system with dynamic precision and bounded neighbor expansion for sub-second location queries
- Implemented real-time messaging with atomic sequence counters and Firestore transactions, ensuring strict ordering despite concurrent writes
- Architected for scale: minimized Firestore read amplification while maintaining consistent message delivery
Stack: TypeScript, React, Firestore, Geohash Indexing
Transformer-based Twitter hate speech detection achieving 96.9% accuracy
- Built an end-to-end NLP pipeline (25K tweets) with text normalization, stratified sampling, and TF-IDF + Logistic Regression baselines achieving 96.0% accuracy and 0.961 weighted F1
- Fine-tuned RoBERTa-base with oversampling and focal loss, boosting macro-F1 from 0.794 → 0.811 and outperforming classical baselines
- Implemented robust evaluation with stratified cross-validation and class-weighted metrics for imbalanced data
Stack: Python, PyTorch, Hugging Face Transformers, RoBERTa, scikit-learn
ResNet-50 CNN classifying fresh vs. rotten fruit with 93% accuracy
- Fine-tuned ResNet-50 on an 80K-image FruitVision dataset with targeted data augmentation and Grad-CAM validation, achieving 93% accuracy and F1 > 0.90 across five fruit types
- Engineered a GPU-accelerated training pipeline with stratified splitting, corruption detection, and adaptive LR scheduling
- Reduced overfitting and increased per-class recall by 18% compared to baseline through systematic hyperparameter tuning
Stack: Python, PyTorch, ResNet-50, Grad-CAM, NumPy, Matplotlib
| Category | Technologies |
|---|---|
| Languages | Python, TypeScript, JavaScript (ES6+), Java, SQL |
| AI / ML | PyTorch, TensorFlow, scikit-learn, Hugging Face Transformers, LangChain, RAG |
| ML Techniques | Fine-Tuning (LoRA), Transfer Learning, Reinforcement Learning (PPO, TD(λ)), Feature Engineering |
| Models | RoBERTa, ResNet-50, Prophet, ARIMA, LLMs |
| Data & Scientific | NumPy, Pandas, Matplotlib, Seaborn, Data Pipelines |
| Backend | Node.js, Express, FastAPI, Flask, Django, GraphQL, REST APIs |
| Data & Infra | Kafka, Redis, PostgreSQL, MongoDB, DynamoDB, Firestore |
| Cloud & MLOps | AWS (Lambda, EC2, S3, API Gateway), Docker, Kubernetes, GitHub Actions, CI/CD, GPU Acceleration (CUDA) |
| Frontend | React, Next.js, Tailwind CSS |
Software Engineer Intern (ML) · UVA Biocomplexity Institute · Nov 2025 – Present
- Engineered automated weekly forecast submissions with 99.9% uptime for CDC health data
- Built forecasting service delivering <500ms latency predictions across 52 jurisdictions
Software Engineer Intern (AI/ML Systems) · RIIG Technology · Aug 2025 – Dec 2025
- Deployed real-time threat detection microservice, reducing mean time to detect by 40%
- Built Kafka + PostgreSQL pipeline processing 500K+ daily OSINT events
Software Engineer Intern · Innova8 LLC · Jun 2025 – Aug 2025
- Developed W3C Verifiable Credentials API for healthcare/education identity workflows
- Built React/TypeScript dashboard reducing manual reporting overhead by 60%
Software Engineer Intern · SS Technology Consultants · Aug 2024 – Dec 2024
- Shipped 3 production features improving task completion rates by 20%
- Automated data migration workflows, saving 15+ hours/week
University of Virginia · B.S. Computer Science · May 2026
Minors in Data Science & Applied Mathematics · GPA: 3.76 · Dean's List ×6
AI Focal Path · Coursework: Algorithms, Software Engineering, Computer Systems, Databases
Teaching Assistant — Multivariable Calculus & Statistics (200+ students)
Currently seeking new-grad Software Engineer or ML Engineer roles starting Summer 2026.
Interested in: ML/AI systems, backend infrastructure, data platforms, or full-stack product engineering at high-growth startups or established tech companies.
Check out my pinned repos below for code samples.