paras harnagle

Architected BFSI financial document automation and compliance validation, eliminating hours of manual review per batch through Azure Document Intelligence and agentic LLM workflows. Built clinical document intelligence pipelines that reduced clinical data entry effort ~70%, achieved >92% F1 on NER (30+ entity types), and cut clinician review time from ~20 min to ~4 min per document.

Delivered accepted GEN AI MVP architectures for Nestlé, including Copilot-style assistants integrated into Microsoft Teams and SharePoint using Azure OpenAI and Copilot Studio. Built presales/POC solutions across legal contract Q&A, AWS multi-agent CFO text-to-SQL reporting, and an AI voice-calling assistant with Twilio and WebSocket-based low-latency responses.

Integrated ML-driven predictive maintenance, anomaly detection, and telemetry forecasting into the Dell DTIAS platform, deploying models as microservices within Kubernetes. Built ZenML + MLflow CI/CD pipelines for automated model training, evaluation, and deployment, and developed Go-based telemetry automation for Dell IDRACs using Kafka, Telegraf, and Spark.

Led Kafka-based data ingestion for real-time software compliance verification across classical and cloud/IoT environments. Built Rust microservices for high-performance, memory-safe license compliance calculations and automated report generation using Go/Rust, PostgreSQL, gRPC, and Kubernetes.

Built a Go-based lead generator filter by scraping Google Ads and LinkedIn Ads and training an XGBoost lead classification pipeline with confusion-matrix evaluation. Designed round-robin and least-connection load balancers in Go and validated behavior with Docker-based testing.

Deployed a TensorFlow Lite model on an ARM Cortex-M7 microcontroller to predict battery State-of-Health (SoH) from raw sensor data and built an embedded IoT framework for data collection, preprocessing, and inference. Developed real-time multi-location video surveillance backend services using Go/gRPC, OpenCV, and WebRTC, orchestrating scaling with Kubernetes.

Designed and implemented APIs for a microservices framework that generates Kubernetes YAML from structured input. Orchestrated multi-node Kubernetes cluster deployments and validated generated output integrity using Docker and microservices components.

Pursuing an M.Sc in Applied Mathematics at Parul University.

Grade: 67%

Completed a BE (Electronics & Telecommunications) at Sinhgad Institute of Technology (SPPU).

Grade: 72%

Completed Class XII at Army Public School, Pune with a 72% score.