VIEW PROJECTAPPLIED AI SYSTEMS / OPEN SOURCE
I design AI systems that can be inspected, evaluated, and shipped with their limits intact.
Agent infrastructure, LLM evaluation, multimodal reasoning, and local deployment—built from architecture through delivery.
SAMSANKAR
SCROLL TO INSPECT
01 / PROFILE
FROM RESEARCH PROTOTYPE TO DEPENDABLE SYSTEM—THE WORK LIVES IN THE ENGINEERING BETWEEN.
LIVE PRODUCT CAPTURE / JUL 2026I am a final-year Computer Science and Engineering student focused on the layer where model behavior meets permissions, evaluation, deployment, and documentation.
Every project here links back to inspectable evidence: source, deployed interfaces, model releases, or clearly labelled architecture derived from the repository.
More about Sam ↗BUILD FOR THE MOMENT AFTER THE DEMO: WHEN SOMEONE NEEDS TO VERIFY, REPRODUCE, OR CHANGE THE SYSTEM.
143typed MCP tools
392node definitions
1.09Mevaluation examples
38datasets
9reward components
4permission modes
143typed MCP tools
392node definitions
1.09Mevaluation examples
38datasets
9reward components
4permission modes
02 / SELECTED SYSTEMS
WORKS® 25—26
A curated set of open-source systems spanning agent orchestration, grounding evaluation, multimodal reasoning, and schema-driven generation.
01 / Agentic tool infrastructure / 2026
MATERIAL PILOT
A local-first MCP bridge that lets AI agents inspect, create, validate, and export procedural material graphs.
SOURCE ↗EVIDENCE / 01
MaterialPilot's public GitHub repository, including the package structure and documented project scope.
Typed operations exposed to compatible AI clients.
Generated procedural-node definitions available to the bridge.
Read-only, assisted, workspace-autonomous, and developer.
The bridge operates without external data transmission.
02 / LLM grounding & evaluation / 2026
HALLUCINATION GUARD
An OpenEnv-compatible training and evaluation environment for factual grounding, citation discipline, calibration, and refusal behavior.
SOURCE ↗EVIDENCE / 01
Live Space overview with documented dataset scale, task tiers, reward components, and environment workflow.
Examples integrated into the training and evaluation environment.
Datasets unified through the environment.
Signals combined into the reward design.
Eight hallucination categories across five severity levels.
03 / Multimodal reasoning / 2026
AEGIS
An interpretable deepfake-analysis pipeline using staged visual reasoning, self-critique, and contradiction-aware evidence aggregation.
SOURCE ↗EVIDENCE / 01
The published AEGIS Kaggle notebook, with the staged deepfake-analysis workflow available for inspection.
Visionary, Inquisitor, and Self-Critique stages.
Two fine-tuning approaches released for inspection.
A quantized release intended for lighter local inference.
A contradiction-aware cross-encoder stage before the verdict.
06 / ACTIVE R&D
EXPERIMENTAL / SOURCE MEDIA NOT SUPPLIED
PROMPTLESS GENERATOR
A schema-driven application generator that converts database structure into a functional initial interface without requiring a natural-language prompt.
Read the R&D note ↗07 / CAPABILITY MATRIX
BUILT ACROSS THE FULL SYSTEM.
Select a capability to trace where it appears in the work.
SELECTED CAPABILITYMCP
Used in: MaterialPilot · HallucinationGuard
01 / Model engineering
02 / Agentic systems
03 / Infrastructure
04 / Applied AI
Inspectability
Intermediate evidence stays visible.
Local-first
Authority and data stay close to the user.
Reproducibility
Systems are documented, testable, and repeatable.
Clear limits
Unknowns are named instead of designed away.
08 / EXPERIENCE TRACE
FROM INDEPENDENT R&D TO DELIVERY.
INDEPENDENT R&D / MAR—JUL
AI Infrastructure & MLOps Engineer
Built three open-source AI systems across architecture, fine-tuning, evaluation, deployment, testing, and technical documentation.
IT LINKS / INTERNSHIP / 10 DAYS
Full-Stack Web Development Intern
Delivered a bulk certificate-generation application from requirements through testing, debugging, and handoff.
09 / EDUCATION
Bachelor of Engineering
Computer Science and Engineering
AAA College of Engineering and Technology2023—2027 expectedCGPA 7.8 / 10
10 / CERTIFICATION ARCHIVE
Oracle
OCI 2025 Certified AI Foundations Associate
Oracle
OCI 2025 Certified Foundations Associate
NPTEL
GPU Architectures and Programming
NPTEL
CPU Architecture
NPTEL
Industry 4.0 and Industrial Internet of Things
Google Analytics Certification
MongoDB University
Connecting to MongoDB in Node.js
11 / START A CONVERSATION