Local-first where it matters
Keep authority and sensitive data close to the user.
Between research prototypesand dependable AI products.
SAM SANKAR / AI SYSTEMS ENGINEER
I am a final-year Computer Science and Engineering student focused on evaluation, tooling, deployment, and system design.
The work turns difficult AI workflows into inspectable systems: typed agent tools, multi-signal evaluation environments, and multimodal pipelines with visible reasoning stages.
THE PERSON BEHIND THE SYSTEMS
Curiosity, sharpened into engineering practice.
I like systems that make their own logic legible. That means treating evaluation, permissions, data flow, deployment, and failure states as product decisions—not afterthoughts.
The visual language follows the same idea: precise structure, warm human context, and just enough motion to make the work feel alive without hiding the evidence.
CURRENT FOCUS
Open-source infrastructure for agents, evaluation, and local inference.
I care about the seam between model behavior and engineering reality: permissions, revision safety, datasets, reward design, quantization, containers, APIs, and documentation.
WORKING PRINCIPLES / 001—005
Inspectability over opacity
Expose state, evidence, revisions, and failure boundaries.
Reproducibility over demos
Document the path from environment to result.
Evaluation over vibes
Use explicit signals, assumptions, and measurable evidence.
Clear limits over inflated claims
Credibility grows when constraints stay visible.
AVAILABILITY