AI Engineer @ Smart Government Technology LLC (Pulscribe.ai) | LLM Optimization & Production Deployment
I am an AI/ML Engineer specializing in Generative AI, LLM Inference Optimization, and Production Deployment. Currently, I work remotely at Smart Government Technology LLC (Virginia, USA), architecting a comprehensive, HIPAA-compliant hospital management platform that integrates AI-powered medical transcription, EHR/EMR sync, and automated billing workflows.
- 🎓 B.Tech in AI & ML at CHARUSAT (CGPA: 9.12)
- ⚡ Focus Areas: Scalable Backend Microservices (FastAPI, K8s), LLM inference (vLLM, GPTQ, Marlin kernels), Medical AI (ASR, NLP, RAG)
- 🌱 Exploring: AI Agents, Multi-Modal Medical AI, Regional Language TTS
- 💡 Mission: Democratizing AI for healthcare and making advanced models accessible on consumer hardware.
AI Engineer | Smart Government Technology LLC (Remote, Virginia, USA) Building an all-in-one, HIPAA-compliant hospital management platform
- Architected FastAPI backend services with robust system design, database modeling, and strict HIPAA/PHI compliance.
- Implemented an AI-powered medical scribe utilizing ASR to transcribe and structure doctor-patient encounters automatically.
- Integrated ICD-10 code automation for seamless insurance processing and billing claim tracking.
- Orchestrated production workloads using Redis, Celery, and Prometheus within a Kubernetes microservices architecture.
AI/ML Intern | Smart Government Technology LLC
- Advanced Medical Domain ASR models for clinical transcription, building end-to-end audio pipelines with noise reduction and domain-specific fine-tuning.
- Collaborated on early system architecture decisions for the core hospital management platform.
AI/ML Intern | Aiginno Pvt. Ltd.
- Solely built the MVP of Ginnobot.ai (a no-code chatbot builder), handling end-to-end product development including DB design, backend APIs, and frontend UI.
- Developed custom n8n plugins to enable seamless workflow automation directly within the platform.
Freelance AI & Full-Stack Engineer | Upwork
- Built and shipped WangúCRM, a WhatsApp-based CRM SaaS platform, owning the backend architecture, API integrations, and production deployment.
- Fine-tuned an Arabic ASR model (data preprocessing, segmentation, evaluation), delivering the project with a 5-star client rating.
| Domain | Technologies |
|---|---|
| GenAI & LLMs | vLLM GPTQ/AWQ Llama.cpp Ollama LangChain RAG Marlin Kernels |
| AI / ML / DL | PyTorch TensorFlow Transformers Hugging Face YOLOv8 Whisper F5-TTS |
| Backend & APIs | FastAPI Flask Node.js Express AsyncIO Uvicorn GraphQL |
| DevOps & Cloud | Docker Kubernetes Redis Celery Prometheus AWS GCP CI/CD |
| Languages | Python TypeScript JavaScript SQL Bash |
| Databases | PostgreSQL MongoDB MySQL Firebase |
🏥 Pulscribe.ai – AI-Powered Hospital Management Platform
Comprehensive, HIPAA-compliant system integrating patient intake, EHR/EMR sync, and billing workflows.
- Developed an AI medical scribe that automatically transcribes and structures clinical encounters.
- Implemented ICD-10 code integration for automated insurance processing and billing.
- Architected microservices with FastAPI, Kubernetes, Redis, and Prometheus for high availability.
🗣️ F5-TTS: Gujarati Text-to-Speech Model
Fine-tuned TTS model using the IIT Madras Gujarati speech dataset to make AI inclusive for regional Indian languages.
- Trained for 30–40 hours on GPU; achieved WER ~15%, CER ~20%.
- Currently co-authoring a research paper and preparing for public deployment on Hugging Face.
🎬 SnapVid AI – AI-Powered Video Generation Platform
🏆 1st Place Winner – 2024 Solution Challenge (IT-CSPIT) | Prize: ₹7,500
- Transforms text prompts and music into beat-synced videos using Kling AI, Librosa, and MoviePy.
🤖 WangúCRM – WhatsApp-Based CRM SaaS
Built end-to-end for an Upwork client — from product planning to production deployment.
- Integrated WhatsApp Business API for automated customer engagement and message routing.
- 📄 Research Publication: "Enhancing Human Detection and Counting Efficiency Through Deep Learning" — Published in AIP Conference Proceedings (ICRAIC 2024, Scopus & Web of Science indexed).
- 🥇 1st Place Winner – 2024 Solution Challenge (IT-CSPIT) for SnapVid AI.
- 🌐 Open Source: Published quantized MedGemma 27B GPTQ model on Hugging Face, optimizing high-throughput medical inference (reduced VRAM from 54GB to 15GB).
- 🎓 Academic Excellence: Graduated with a 9.12 CGPA in B.Tech AI & ML at CHARUSAT.
B.Tech in Artificial Intelligence & Machine Learning Charotar University of Science and Technology (CHARUSAT) | 2022 – 2026
- CGPA: 9.12 / 10.0
- Final Year Project: AI Based Healthcare NLP System for Clinical Speech Recognition and SOAP notes Generation
I'm always open to discussing new AI projects, LLM optimization techniques, or innovative backend architectures.
"AI is not just a tool; it's a bridge to innovation, transforming ideas into reality and shaping a smarter future."