Skip to content
View Gryffin9's full-sized avatar
2 followers · 1 following

Achievements

Achievement: Pair Extraordinairex4Achievement: Pull Sharkx3Achievement: QuickdrawAchievement: YOLO

Achievements

Achievement: Pair Extraordinairex4Achievement: Pull Sharkx3Achievement: QuickdrawAchievement: YOLO

Highlights

  • Pro

Block or report Gryffin9

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Gryffin9/README.md

Aman Panda

AI Product Engineer · Agentic Systems · Full-Stack Product Engineering

I build production software and agentic systems where generated work has to survive independent evaluation, deterministic validation and real-world failure modes.

Founding Product Engineer & Technical Lead at Clymber · Incoming MPhil in Scientific Computing (High-Performance Computing), University of Cambridge.

Email · LinkedIn · Engineering case studies

Systems map: build with explicit intent, verify independently, measure persisted evidence, learn through reviewed constraints. A bounded repair path returns from verify to build; learned constraints shape the next build.

Most production work lives in private repositories; this profile exposes only non-sensitive engineering evidence.

Selected engineering evidence

Production systems: 0 to 265 automated test files (snapshot 2026-09-10); recorded local unchanged-content startup benchmark 5.317 seconds to 155 milliseconds.

Production: 0 → 265 test files; 5.317 s → 155 ms recorded local unchanged-content startup benchmark.

Agentic reasoning: 216 to 1,896 passing verification tests (snapshot 2026-09-10); independent evaluation, deterministic gates, persisted learning loop.

Reasoning: 216 → 1,896 recorded passing tests. Independent evaluation and accumulated failure-mode checks; test count is not a correctness guarantee.

Video quality gates: 2,486 raw findings, then 40 genuine defects after detector calibration, then zero blockers after content correction.

Video: 2,486 raw findings → 40 genuine defects → 0 blockers. Detector calibration first; content correction second.

Test-count snapshot: 2026-09-10. Methodology · Refresh

How I engineer with agents

Explicit task contracts, isolated implementation, deterministic validation, regression evidence and human acceptance. Run the synthetic contract example →

Scientific computing: Electrostrictive metamaterial simulation and validation data, alongside my incoming Cambridge MPhil.

Building now: a local-first, evidence-traceable research workflow with versioned approvals and persisted recovery.

Pinned Loading

  1. agentic-engineering-template agentic-engineering-template Public

    Executable template for AI-assisted engineering with task contracts, isolated workflows, validation gates, CI, and human review.

    Python

  2. electrostrictive_metamaterial_study_data_files electrostrictive_metamaterial_study_data_files Public

    Simulation and validation data supporting the electrostrictive metamaterial preprint (doi:10.5281/zenodo.18450271): COMSOL sweeps, field exports, optimization traces.

  3. engineering-case-studies engineering-case-studies Public

    Measured engineering case studies across production systems, agentic evaluation, and scientific computing.

    Python