Software Engineer · GenAI Developer · Distributed Systems
MSc Computer Science @ Memorial University of Newfoundland, Canada 🍁
Previously Software Engineer @ Techolution · Presented at New York AI Summit 2023 · Published in IEEE
I build production-grade AI-powered systems end-to-end — from Kafka-backed ingestion pipelines and vector search to user-facing APIs and interfaces. I care about systems that are fast, observable, and built to scale.
- 🔭 Currently building a distributed RAG pipeline with Kafka + Qdrant + OpenTelemetry
- 🏥 Developing a YAML-driven clinical data platform and contract/invoice generation pipeline @ CAIR
- 🤖 Built AI SaaS platforms processing enterprise data at scale (100K concurrent users)
- 📄 First-author IEEE publication on deep learning-based product classification
Production-grade distributed RAG system built from scratch. Kafka (4-partition) routes documents through parallel PDF/HTML workers → adaptive chunker (fixed/semantic/hierarchical) → swappable embeddings (HuggingFace / OpenAI / Ollama) → Qdrant vector store + PostgreSQL metadata. Fully instrumented with OpenTelemetry + Jaeger tracing and Prometheus metrics. CI/CD via GitHub Actions with Docker Compose integration tests on every commit.
Python FastAPI Kafka Qdrant PostgreSQL OpenTelemetry Prometheus Docker GitHub Actions
ML-driven system that predicts invocation patterns and triggers pre-warming before demand spikes — achieving an 83% reduction in cold start frequency. Real-time metrics pipeline with adaptive thresholding feeds live system signals into the prediction engine.
Python Scikit-learn AWS Lambda Real-time Metrics
Full-stack AI platform extracting, classifying, and structuring financial transactions from unstructured documents — cutting manual data entry by 70%. Extraction and classification pipelines normalise raw documents into analytics-ready data.
Next.js FastAPI LLMs Document AI PostgreSQL
Cloud-native file management platform with secure authentication, role-based access, and usage analytics. Optimised UX flows lifted onboarding conversion by 40%.
Next.js Appwrite TypeScript TailwindCSS
| 🏥 CAIR, MUN | YAML-driven form builder · auto-generates Django models, REST APIs, UI forms · Contract & Invoice generation pipeline |
| 🤖 Techolution | CDN-powered AI chatbot (100K concurrent / 250K peak) · Micro frontend ETL platform · AI SaaS @ NY AI Summit 2023 · CoPilot (60% accuracy boost) |

