AI/ML Engineer and Full-Stack Developer with hands-on experience building and shipping production-grade machine learning systems, intelligent backend APIs, and MLOps pipelines. Proficient in Python, FastAPI, Scikit-learn, MLflow, PySpark, and React — with a proven ability to take models from prototype to fully monitored, cloud-deployed production. Experienced in vector database integration, RAG architectures, LLM orchestration, and automated retraining workflows. Strong communicator who translates complex ML outputs into actionable, business-relevant solutions.
Languages & Frameworks
Python, SQL, PySpark, FastAPI, SQLAlchemy, React/Vite
Machine Learning
Scikit-learn, GaussianNB, KMeans, Random Forest, PCA, TF-IDF, Isolation Forest, HuggingFace Embeddings
MLOps & Pipelines
MLflow, APScheduler (auto-retraining), Prometheus, Evidently AI drift detection, GitHub Actions CI/CD
AI & LLM Integration
Claude API, Gemini Flash, RAG architecture, Milvus vector DB, section-aware document chunking
Databases & Storage
PostgreSQL, Milvus (384-dim cosine similarity), Alembic migrations, Pydantic schema validation
Cloud & DevOps
AWS EC2, Docker, GitHub Actions, containerized deployment, Databricks Lakehouse patterns
Frontend
Tailwind CSS, Lucide, Chart.js (radar/donut), WebSocket streaming, responsive design
Languages
English, Malayalam, Hindi, German
Independent · Thiruvananthapuram, India
- Designed and shipped multiple end-to-end ML and AI systems, managing full lifecycle from architecture to cloud deployment.
- Built a production MLOps platform with automated retraining, MLflow versioning, Prometheus monitoring, and GitHub Actions CI/CD deployed on AWS EC2.
- Developed a Medical RAG backend integrating Milvus, HuggingFace embeddings, Claude API, and PostgreSQL — 8 production REST endpoints with full observability.
- Engineered a real-time inference dashboard supporting three concurrent ML models with WebSocket streaming and Evidently AI drift detection.
- Delivered full-stack web applications using React/Vite/Tailwind, translating business requirements into polished, production-quality frontends.
Devaja Traders
- Managed financial data pipelines for inventory, billing, and reconciliation; developed structured reporting workflows reducing manual effort by 40%.
- Applied data analysis to identify cost variances and operational inefficiencies, translating findings into actionable business recommendations.
→ bensbasil.in/quiz
- Decoupled ML training pipeline (feature engineering, training, serving); APScheduler automated retraining with drift triggers; MLflow experiment tracking and model registry.
- Full Prometheus instrumentation — inference latency, prediction drift, request volume; zero-downtime CI/CD via Docker + GitHub Actions on AWS EC2.
- Extended classifier to predict stress-type segments alongside DISC profiles; resolved Pydantic schema gap suppressing model confidence stats (now 91%).
PythonScikit-learnMLflow
PrometheusAPSchedulerDocker
GitHub ActionsAWS EC2
→ bensbasil.in/medical-rag-app
- Production-grade RAG backend with 8 REST endpoints; Milvus vector DB with cosine similarity search and 384-dim HuggingFace embeddings over PostgreSQL metadata layer.
- Section-aware PDF chunking with source attribution and confidence scores; multi-model backend switching between Claude API and Gemini Flash at runtime.
- Alembic-managed schema migrations; zero manual SQL patches across all environment deployments.
FastAPIMilvusClaude API
PostgreSQLHuggingFaceSQLAlchemy
AlembicReact
- Three concurrent ML models (Random Forest, TF-IDF+LR, Isolation Forest) with WebSocket streaming, real-time Prometheus metrics, and Evidently AI drift detection.
- Query log analysis and inference pipeline profiling to identify and resolve throughput bottlenecks in high-frequency prediction workloads.
PythonWebSocketsPrometheus
Evidently AIReactRecharts
Team project · Frontend contribution
- Built the frontend dashboard with Chart.js radar/donut visualizations and an AI feedback panel with model-toggle support across Hugging Face and Gemini backends.
ReactChart.jsGemini API
HuggingFace
Education & Certifications
Bachelor of Computer Applications (BCA)
IGNOU
2023 – 2025 · CGPA 7.97/10
Google Advanced Data Analytics
Google / Coursera
2025 · Professional Certificate
12th — CBSE
Kendriya Vidyalaya Pangode
Thiruvananthapuram