Artificial Intelligence Engineer | AI Automation Expert | Technical Architect
Email | LinkedIn | Portfolio (Google Drive)
Google Certified TensorFlow Developer (valid till July 2026)
Top 5% AI Professional (Pro5 Certificate, 2025)

I'm a high-impact, self-taught builder who turns complex, real-world problems—especially in clinical AI, workflow automation, and digital dentistry—into production-ready systems, fast.

• 🚀 Delivered 100+ personal and professional projects (CV, NLP, LLMs, Workflow Automation) from scratch in an early-stage startup environment, with minimal guidance, resources, or external codebases.
• 💡 Thrive in “zero-to-one” situations: take full ownership, build fast, iterate independently, and solve hard problems in ambiguity.
• 🔬 Currently working on Machine Learning, Deep Learning, and LLM agent ecosystems
• 🧠 Google Certified TensorFlow Developer
• 🏅 Top 5% AI Professional (Pro5, 2025)
• 🤝 Passionate about hands-on AI, automation, and building full-stack, end-to-end solutions
• 📈 Always learning, always iterating—moving fast and raising the bar in every project

AI / ML: YOLOv8, nnUNet, PointNet++, DGCNN, PyTorch, embeddings, RAG, LLMs (GPT‑4, Mistral)
Automation / Agents: n8n, Retell AI, Twilio, Telegram/Gmail APIs, human-in-the-loop orchestration
Dev / Cloud: Python, React, FastAPI, MongoDB, MySQL, Docker, GitHub Actions, AWS (Lambda, S3, API Gateway)
Medical Imaging: DICOM/NIfTI, Slicer, Pulpy3D, Dental Segmentator
3D Workflows: Blender (bpy, bmesh), STL processing, CAD/CAM scripting

I turn complex, messy ideas into production-ready AI/ML systems.
From hands-on hardware prototyping and 3D modeling to LLM/AI agent orchestration and cloud deployment, I deliver solutions that drive clinical, business, and workflow value.

• Independently delivered full-stack, cloud-native solutions in fast-paced, resource-limited startup settings
• Fast learner, creative problem-solver—thrive in data-scarce and resource-limited environments
• Built automation systems, advanced computer vision models, intelligent voice/chat agents, and cloud-first software
• Product owner and architect for next-generation dental patient management systems, imaging, and agent automation

My AI-driven workflow automation ecosystem is built to eliminate repetitive admin tasks and boost productivity for clinics and professionals.
All designed, built, and deployed as a solo engineer in a lean, early-stage startup context.

Key highlights:

• Multi-agent Automation (n8n, LLMs, Retell AI, Twilio, Google APIs):
  • Appointment Management Agents: Voice/chat agents handle scheduling, reminders, and follow-ups—integrated with Google Calendar, Telegram, and WhatsApp.
  • Sales Campaign Agents with Human-in-the-loop Feedback: Run multi-stage outreach, collect feedback, and escalate for human review/approval.
  • Custom Receptionist Agent: Handles inbound and outbound calls using Retell AI and Twilio; can transfer calls, escalate complex cases, and provide a natural conversational experience.
  • Personal Telegram Bot (Full Workflow Automation): My own custom bot manages emails, schedules meetings, sends automated emails, conducts research, and interacts with external APIs—all from Telegram chat.
  • Cross-Agent Communication: Agents delegate tasks to each other, with supervisor logic and human oversight for quality and safety.
  • Automated Marketing, Reminders, and Documentation: Automates messages, sends proposals, and syncs patient/staff data.
• Impact: Achieved 80%+ admin automation, sub-minute response times, and robust reliability with human oversight.

(All projects were self-driven, from zero-to-one, often without any pre-existing codebase or external support, and validated by clinical/end-user feedback.)

Built an AI model that outperforms experts at identifying implant make/manufacturer using synthetic wax phantoms to address data scarcity. Achieved mAP@0.5 = 0.99, mAP@0.5-0.95 0.813 and Precision: 0.968.

Developed a Blender-Python pipeline that calculates centroids and normals to programmatically align abutments on STL jaw scans—validated via clinical 3D-printed fit testing.

Constructed an STL-to-point-cloud pipeline, manually labeled data, and trained PointNet++/DGCNN models with ~97% accuracy for implant base identification.

Designed laser-engraved, geometrically unique collars for robust orientation detection in scans. Prototyped and validated via patent analysis and scanning tests.

Led product vision and architecture for a cloud-native, voice-enabled PMS: voice-driven EHR entry, AI assistants, automated billing, secure multi-role access, powered by React, AWS, APIs, and AI.

Built full-stack ML pipelines using nnUNet, MONAI, DICOM/NIfTI data, and Slicer to generate 3D meshes for clinical workflows. Demonstrated high-quality pulp segmentation from Pulpy3D.

Created a multi-agent system using n8n, LLMs (OpenAI/Mistral), Retell AI, Twilio, and Telegram to automate clinic workflows—achieved over 80% automation and sub-minute task execution.

View sample projects, code, diagrams, and demos:
➡️ Google Drive Portfolio

• Google Certified TensorFlow Developer (valid till July 2026)
• Top 5% AI Professional (Pro5 Certificate, 2025)
  For successfully completing the Pro5 AI Interview for AI Engineer — August 2025
• Built solutions recognized for outperforming manual expert workflows in clinical settings
• Proven track record of self-driven, high-impact innovation in lean, ambiguous, and high-pressure environments

• LinkedIn: https://www.linkedin.com/in/jayavathsan/
• Email: jayavathsan@gmail.com

Let’s build the future of intelligent automation—together!