Scientific software / AI systems / open source

I build systems that make the hard parts faster, safer, and useful.

Abid Khan is a 19-year-old Computer Engineering student in India, accepted for GSoC 2026 with NumFOCUS/scikit-bio. His work spans Numba-optimized bioinformatics, AI reliability tooling, civic tech, healthcare AI, and live developer platforms.

Accepted GSoC contributor Live users + OSS reviews Benchmarks, not vibes
GSoC 2026NumFOCUS / scikit-bio
500+ usersAdventurers Guild
99 reposGitHub public footprint
NASASpace Apps Regional Finalist
Open PawsAI + advocacy tooling

01 / Case Files

Flagship work, presented as systems.

The strongest projects here are not just apps. They are operating environments: scientific compute paths, user marketplaces, civic intelligence workflows, and clinical AI infrastructure.

The design language is intentionally light, precise, and evidence-led: closer to a product/research workspace than a flashy student landing page.

Case 02 / 04

Developer marketplace

Adventurers Guild

Quests → XP → Rank → Work

Gamified platform where developers complete real quests, gain XP, and rank up from F to S tier. Selected for GirlScript Summer of Code 2025 and NSOC, with active user and contributor workflows.

500+Users
25Forks
LiveStatus
Next.jsTypeScriptPrismaStripe
Case 03 / 04

Civic AI

Open Permit

Permit → Legal context → Objection

AI system for generating legal objections to factory farm and planning permit proposals across jurisdictions. Born at Code 4 Compassion, then developed into a civic intelligence platform.

8+Countries
37Frameworks
LiveStatus
AI/NLPCivic TechLegal Workflows
Case 04 / 04

Healthcare AI

Knight Medicare

Clinical data → Safety → Backend

Clinical AI infrastructure for diagnosis support and patient data management. Abid is CTO and co-founder, building production readiness and security hardening into the backend.

CTORole
AIClinical tooling
OpenPR #206
Full-stackSecurityBackendHealthcare

02 / Evidence

Technical proof beats portfolio theater.

The site should make the work legible: benchmarks, reviewed PRs, production risk reduction, and concrete engineering constraints.

If a claim here isn't attached to a PR, a benchmark, or a commit hash, it doesn't belong on this page.

Selected OSS contributions

RepoWorkSignal
scikit-bio#2429, #2436Array API foundation for GSoC performance work
daytona#4107Merged sync-session stdin handling fix
nextflow#6869Clarified staged path behavior in bioinformatics pipelines
scanpy#3986Reduced duplicated scientific docs drift
codecarbon#1080CLI robustness for emissions tooling
kepler#2431Minimal install path for Kubernetes energy monitoring
benchmark.snapshot
PERMANOVA row-tile Numba kernel (merged, PR #2488)

~8x faster than single-permutation Cython
~2.4x off the hand-optimized C++ reference
at n=25,000 / 9,999 permutations
Accuracy: max abs diff < 1e-15 vs Cython

GPU-resident DistanceMatrix (merged, PR #2499)
PERMANOVA + Mantel on the Array API (in review, PR #2503)
NVIDIA RTX 4060 (CuPy) .......... matches reference
AMD Instinct MI210 (PyTorch) .... matches reference

Separate GPU kernel spike, not yet a PR:
same PERMANOVA kernel source ran correct
on RTX 4060, MI210, and MI300X

dispatch: Numba CPU -> Array API (shipped) -> Numba CUDA/HIP (next)

03 / GitHub

Recent work clusters around trust.

The current engineering pattern is clear: AI output validation, security hardening, maintainership structure, and production readiness.

Recent public PRs

scikit-bio: GPU-capable DistanceMatrix

Array API support so DistanceMatrix can hold a GPU-resident buffer, verified on an AMD Instinct MI210.

Merged

scikit-bio: GPU support for PERMANOVA & Mantel

Array API compute path for both statistics on a GPU-resident matrix, same semantics as NumPy, verified on an AMD Instinct MI210.

In review

desloppify intelligence gate

De-hallucination gate for API veracity in Open Paws AI workflows.

Merged

Adventurers Guild security

TLS production guard, PII log removal, and URL sanitization.

Open

Knight Medicare readiness

Security hardening and backend production enhancements.

Open

pydantic-ai validation

Strict runtime validation for tool outputs in Python AI tooling.

Open

Current roles

GSoC Contributor

NumFOCUS / scikit-bio. CPU Numba engine and GPU-resident DistanceMatrix merged; PERMANOVA/Mantel GPU support in review.

In progress

Rapid Impact Developer

Open Paws AI tooling for animal advocacy and campaign workflows.

Active

CTO & Co-Founder

Knight Medicare, building clinical AI and backend infrastructure.

Building

Google Gemini Campus Ambassador

Representing Gemini and AI education on campus.

Campus

04 / Field Notes

The blog is where the proof gets unpacked.

The homepage gives the signal. The blog gives the architecture, benchmarks, tradeoffs, and operating notes behind the work.

05 / Contact

Reach out when the work has teeth.

Best fit: scientific Python, AI reliability, civic tech, animal advocacy tooling, healthcare AI, and serious open-source collaboration.

"Quick call to pick your brain" gets a slower reply than a PR link or a repro case.

Identity

NameAbid Khan / Abi
Age19
EducationComputer Engineering, GTU - GIDC Navsari
LocationIndia, UTC+5:30

Links

Emailabid@openpaws.ai
Email
GitHubgithub.com/LarytheLord
Open
LinkedInlinkedin.com/in/abikhn
Open
Xx.com/larysonlawliet
Open
Instagram@aulterabid
Open