About

Hi, I'm Suyash.

I'm an AI engineer and full stack developer. I build LLM systems with guardrails and evals, and I do forward deployed work: sit with the people using the system, ship what they need, and leave something the next person can maintain.

Bio

What I do

I build LLM systems end to end: guardrails, evaluation pipelines, and the application code around them. I test against live providers and write evals that say what they cannot claim.

At Anterra Corp I fine-tuned YOLOv8m and integrated BoT-SORT tracking on Jetson Orin Nano edge hardware. I built a custom Nepali Devanagari ANPR model and demonstrated the traffic violation detection system to senior police and ministry officials.

On freelance projects I work directly with the people using the software. I have sat with factory workers to refine inventory workflows and with school administrators and teachers to fix bugs in an academic management system.

Honors

Recognition.

World 2nd

Cambridge International A-Level Computer Science, also Nepal Top

1540

SAT, 99th percentile

A* A*

Computer Science and Mathematics A-Levels

1,500+

students reached through WIZ, the computer literacy nonprofit I founded

  • Best Academic Performance Award and AS Level Topper at Budhanilkantha School
  • National Economics Olympiad Finalist
  • Best Delegate twice at BNKS Model UN

Experience

Where I've worked.

Jul 2026 to Sep 2026

Lead Engineer, Flamide

Designed and developed the core architecture of an AI-native CRM using Next.js, Supabase, and DeepSeek, with multi-tenant data isolation, real-time business-data grounding, AI guardrails, and cost- and latency-optimized model routing.

Aug 2025 to Jul 2026

Co-Founder and Lead AI Engineer, Anterra Corp (Karma)

Built an AI-powered traffic violation detection system by fine-tuning YOLOv8m and integrating BoT-SORT tracking on Jetson Orin Nano edge hardware. Developed a custom Nepali Devanagari ANPR model and demonstrated the system to senior police and ministry officials.

2024 to 2026

Founder, World of IT Zealots (WIZ)

Founded and led a computer literacy initiative reaching 1,500+ students across underserved schools. Developed a self-directed digital literacy curriculum, delivered training to 600+ students, and built AI integrated interactive modules with custom harness and context optimizations.

2022 to now

Freelance Web Developer

Built a bank credit reward system with event-driven Java/Spring Boot and Python architecture, a factory operations and inventory management system for real-time tracking and production scheduling, and an academic and administrative management system for student progress and teacher-facing workflows.

Oct 2024 to Dec 2024

Intern, IM Swedish Development Partner

Researched policy implementation gaps affecting freed Kamaiyas and Kamlaris. Revamped the Freed Kamalari Development Forum website, improved digital outreach materials, and developed a crowdfunding strategy with donation buttons, QR codes, and international donor engagement.

Research and education

Background.

Independent research

Testing the Platonic Representation Hypothesis at toy scale

Reproduced a reference representation-alignment method from paper descriptions and matched the published result, 96.37% vs 96.01%. Recovered unsupervised alignment from a null result to 77.3% through preprocessing corrections and a second unsupervised method.

Education

Budhanilkantha School, Cambridge International A-Levels

Computer Science A*, Mathematics A*, Economics A, Business A, English General Paper A.

2025

Certifications, DeepLearning.AI and Stanford

Advanced Learning Algorithms and Supervised Machine Learning: Regression and Classification.

Sep 2025

Research Assistant, Access Planet Nepal

Co-authored a publication in Sambahak: Human Rights Journal.

Skills

What I reach for.

AI and ML

  • Python, PyTorch
  • YOLOv8, BoT-SORT
  • Computer vision, ANPR
  • LLM evaluation pipelines
  • AI guardrails and grounding
  • Model evaluation and validation
  • scikit-learn, pandas, NumPy
  • Representation learning
  • Edge AI deployment
  • Machine learning, model fine-tuning

Backend and data

  • FastAPI, SQLAlchemy
  • Supabase, PostgreSQL
  • Row-level security
  • REST APIs
  • Multi-tenant architecture
  • Java, Spring Boot
  • Real-time data systems
  • Database design and data modeling
  • Authentication and RBAC
  • Node.js, TypeScript

Frontend and tooling

  • Next.js, React
  • HTML and CSS
  • Docker
  • CI/CD, GitHub Actions
  • pytest, automated testing
  • Linux
  • System design
  • API integration
  • Technical documentation
  • JavaScript

Languages

English (fluent), Nepali (native), Hindi (proficient).

Contact

Want to talk through a project?