Swanand Kadam — Senior AI / Full Stack Engineer

Selected Work

Eight projects. Real outcomes.

P.01

Gaslight — Automated Pentest CLI for MCP AI Agents

Physical proof · 17 attacks + 7 auth probes · No API key required · A/B/C/F grading · v0.2.4 on PyPI

Physical proof only — never opinions. A finding is CONFIRMED only when something physically happened: a unique canary token arrived at a local HTTP sink gaslight controls, real out-of-bounds data returned, or a tool's own state contradicted its own description. No second model grades the first. 5 real confirmed findings against a 136-server registry-verified corpus.
17 core attacks + 7 auth probe checks. Core suite: 14 model-free, runs with zero API key. Auth probes attack the HTTP front door (no-credential, alg:none JWT, audience passthrough, cleartext transport) — a separate threat model from the tool layer. OWASP MCP Top 10 — 7 fully covered, 1 partial, 2 out of scope by design.
A/B/C/F grading + auto-discovery. F = proven exploit (data crossed a boundary). C = disclosed but not exfiltrated. B = internal info gap. A = nothing fired. Multi-language auto-discovery (Python/Node/Go/Rust) + interactive wizard + self-healing venv retry means one command, no flags, no guessing. v0.2.4 on PyPI — Trusted Publishing, explicit sdist whitelist guards scan results from accidental publish.

Attack Pipeline — v0.2.4 · Two-Layer Architecture

Auto-Discovery
+ Wizard

py/node/go/rust
.gaslight.json

→

Target
MCP Server

stdio · HTTP
SSE auto-detect

→

17 Attacks
(fresh conn each)

+ SSRF encoding
evasions

→

Canary Sink
port 0 stdlib

physical proof
binary verdict

→

A/B/C/F
+ Blast Radius

severity-tiered
3-ring diagram

Auth Probe Layer — HTTP targets only (front door, before tools)

No credential
honored → F

critical

alg:none JWT
accepted → F

critical

Audience
passthrough → F

critical

Cleartext
transport → C

high

JWT no exp/
audience → B

medium

Model-vs-code 2×2 per tool: GENUINELY DEFENDED · DEFENSE-IN-DEPTH · FRAGILE (no backstop) · CRITICAL. Scope-creep baseline v2 tracks per-tool capability fingerprint (file/net/code) for CI — only flags privilege growth, never shrinkage.

Python asyncio MCP Security Anthropic Ollama pytest

P.02

CodeMas — AI Coding Assessment Platform

Masai School · First engineer reporting to CTO · 10K concurrent users · ~10K submissions/sec

Architecture Migration — migrated execution layer from Redis + Celery + Docker to AWS Lambda + SQS FIFO. Trade-off: lost persistent pub/sub but gained per-invocation isolation, zero host management, and elastic scale at exam deadline bursts. Deterministic isolation without running Docker on a persistent host.
Exam Generator — GPT drafts complete exams (questions + test cases) from topic and difficulty; human reviews before any persistence. ~80% faster exam creation.
Rubric Scoring — async GPT grades every submission on correctness, code quality, approach, and edge cases (0–2 each) with per-dimension justifications. Consistent feedback at scale across 10K+ submissions — zero trainer overhead.
Socratic Hint — for failed submissions, generates a pedagogical nudge pointing at the concept without revealing the answer. Students self-correct without trainer intervention.
Trainer Dashboard — AI-generated cohort summaries and per-student performance narratives. Replaced manual spreadsheet reporting entirely.
Plagiarism Engine — behavioural signals (paste ratio, speed anomaly, tab switches) + TF-IDF cosine similarity. Reframed from "are submissions similar" to "did this student write this." 19× improvement in catching cases.
Business Impact — automation eliminated manual exam oversight, making multi-cohort scale operationally viable. Each cohort = 400 students × ₹3L = $2M+ in revenue per cohort unlocked for Masai.

Submission Flow — Critical Path (p95 < 10s)

Browser

→

Django API

SELECT FOR UPDATE
attempt gate

→

SQS FIFO

burst buffer
exactly-once

→

Lambda

ephemeral sandbox
per-invocation

→

Postgres

result written

Browser

renders verdict

←

Poll REST / 1.5s

GET /submissions/{id}/

←

Postgres

status check

Supporting (off critical path)

Redis

Celery broker
5 async AI tasks

→

Celery Workers

hints · rubric · exam gen
plagiarism explain · summary

Redis

Dashboard cache
TTL 10s

Redis

Plagiarism weights
TTL 60s

Two-phase plagiarism: Phase 1 behavioral O(N) → suspects K · Phase 2 TF-IDF cosine O(K×N). N²=2B at 10K students → reduced to ~5M. 19× improvement in catch rate.

Django AWS Lambda SQS SSE GPT-4o-mini PostgreSQL

System Design AI Features

P.03

the01.dev — Sovereign AI Tutor Platform

Co-founder · Led engineering (2-person team) · Architecture owned end-to-end · 8 deep CS courses

GEPA — Self-Evolving Tutor — reflective prompt evolution: eval SOUL on test set → reflect on weakest cases → propose improved SOUL → re-eval with Pareto gate (avg improves AND no case regresses) → human approves → apply with rollback backup. Every SOUL version is fingerprinted via SHA hash in the audit log. Agent improves from its own runs without unbounded self-modification.
Sovereign RAG Tutor — Hermes runtime — grounding enforced in code, not by model trust: relevance threshold 0.45; below it the model is never called and a deterministic decline fires. Hermes agent with SOUL.md identity, MCP tools, HindSight session memory, EU-AI-Act audit log written before every inference. RAGAS (faithfulness + context precision) evaluates quality using the local sovereign model. Gate prevents generation when retrieval is weak — grounded or silent.
Two-Engine AI Architecture — Hermes (ReAct agent loop) for the open-ended tutor; LangGraph (deterministic workflow with HITL interrupt/resume + MemorySaver checkpointer) for QuizMe and artifact generation. Design rule: agent when the AI decides its next move; workflow when every step is known in advance. No single framework forced on every problem.
Sovereign Inference + GDPR — generation runs on Ollama/vLLM; nothing leaves the machine. Memory transparency endpoint + DELETE erasure. SOUL hash in every audit row makes self-improvement tamper-evident. Compliance-ready — sovereignty, audit, and erasure from day one.

RAG Tutor Pipeline — Grounded · Audited · Sovereign

Question

POST /hermes/chat

→

JWT + Purchase

Firebase auth
Firestore check

→

Audit Row

written BEFORE
inference (EU-AI-Act)

→

Embed

OpenAI
text-embedding-3-small

→

pgvector HNSW

top-4 chunks
cosine similarity

Browser SSE

sources → tokens → done

←

Ollama / vLLM

100% local
sovereign

←

SOUL + Chunks

Hermes system
prompt

←

Score Gate ≥ 0.45

enforced in code
DECLINE if below

Hermes · ReAct Agent

Reason

→

Tool

→

Observe

↩

Open-ended → RAG Tutor

LangGraph · StateGraph

Plan

→

HITL ⏸

→

Grade

Known steps + HITL → QuizMe

GEPA loop (offline): eval SOUL on test set → 3 weakest cases → LLM proposes new SOUL → Pareto gate → human approves → SHA-fingerprinted deploy. Self-improvement that can't auto-update without oversight.

Hermes LangGraph GEPA RAGAS React 19 FastAPI Ollama / vLLM

System Design Multi-Agent RAG Pipeline

P.04

Munshi — Sovereign GST & Trade Compliance Agent

Bharatvarsh Arts · Local-first · Hermes runtime · Human-in-the-loop · ₹5Cr revenue scale

Fuzzy Invoice Reconciliation — AI adjudicates messy purchase invoices against government GSTR-2B records; deterministic Python computes every rupee. Recovers input tax credit (ITC) that exact-matching forfeits.
Sovereign Architecture — GST side fully local-first, nothing leaves the machine. Model only judges ambiguous matches; all tax arithmetic is exact Decimal computation. Privacy-safe for sensitive financial data.
Analytics Data Platform — Snowflake + dbt ELT pipeline unifies 4 disconnected systems (Shopify, GoKwik, NimbusPost, WhatsApp B2B). GPT-mini extracts structured B2B orders from Hinglish/Marathi chat. 247 actionable exceptions surfaced from ~2,100 orders; ₹15.4L revenue reconciled across channels.
Human-in-the-loop + PII Governance — consequential actions require approval; Snowflake Dynamic Data Masking governs customer PII at mart level; AGENT_READER role sees masked data in ad-hoc queries. Agent queries are auditable and PII-safe by default.

Agent Loop — ReAct + MCP Tool Calls + HITL Gate

Owner Query

plain English

→

Reason

Hermes + SOUL.md

→

MCP Tool Call

JSON Schema
validated

→

Observe

result into
context

→

request_approval()

HITL pause before
consequential action

↩ loop back to Reason until answer complete

Fuzzy Invoice Matching — 4-Signal Score

GSTIN match

weight 0.40

Name token ratio

weight 0.35

Amount ±5%

weight 0.15

Date ±7d

weight 0.10

≥ 0.70 → Auto Match (~80%)

0.50–0.69 → Model Judgment

< 0.50 → No Match

Python Decimal for all tax arithmetic — exact base-10, matches GSTN portal to the paisa. Approved verdicts stored in SQLite and reused month-to-month.

Analytics ELT Pipeline — Snowflake + dbt + Agent Tools

Shopify CSV
GoKwik CSV
NimbusPost CSV

4 sources,
no shared ID

→

WhatsApp B2B
→ GPT-mini
extraction

Hinglish/Marathi
→ structured

→

Snowflake RAW
VARIANT

schema-on-read
no brittle DDL

→

dbt staging
(views)

clean · normalize
fix grain

→

dbt marts
(tables · tested)

37 tests green
PII masked

→

7 Agent
Analytics Tools

order_status
exceptions
revenue

Key Marts

xref_order_identity

identity bridge

fct_order

Order-360

b2c_exceptions

247 issues

dim_vendor_contract

SCD Type 2

fct_revenue

₹15.4L unified

"AI decides, deterministic code computes" — LLM extracts WhatsApp price intent; all ₹ arithmetic is SQL. AGENT_READER role masks PII in ad-hoc queries.

Hermes MCP FastAPI Ollama Snowflake dbt Python ELT Docker

System Design

P.05

Trade Compliance Researcher — Multi-Agent Hermes System

Nous Research Hermes runtime · Researcher → Writer workflow · Open-weight & hosted models

Two-agent pipeline — Researcher agent gathers trade-compliance data via MCP tools; Writer agent synthesises findings into structured reports. Memory persists across turns. Collaborative multi-agent workflow on a single runtime.
Model-agnostic — switches between Ollama open-weight models (Qwen, Llama) and hosted cloud endpoints by changing one config line. Zero code changes to swap model providers.
SOUL + config architecture — agent identity and capabilities declared in SOUL.md + config.yaml; no agent plumbing written from scratch. Fully Dockerised.

Two-Agent Pipeline — Choreography, Independent SOULs

Query

HS code · origin
destination

→

Researcher Agent

Hermes + SOUL.md A
search_regulations
fetch_tariff_db
fetch_document

→

Findings

duties · FTAs
restrictions · docs

→

Writer Agent

Hermes + SOUL.md B
zero tools by config
synthesis only

→

Report

citeable
source-attributed

max_iterations=10 circuit breaker on Researcher. Choreography: Writer coupled only to the findings artifact — either agent upgrades independently. Config swap = model or SOUL change with zero code changes.

Hermes MCP Ollama Docker Python

System Design

P.06

Kalaam — India's First Hindi Programming Language

Open Source · TEDx Bangalore · npm · kalaam-core · 500+ monthly users

Multi-language Architecture — 5 languages (Hindi, Marathi, Bengali, Telugu, Odia) via centralized keyword configuration; adding a new language is one map entry, zero parser changes. Infinitely extensible without touching the interpreter core.
Fully Offline PWA — zero API calls, service-worker cached, runs in the browser on a budget Android phone with no internet. Reaches tier-3 city students locked out of every other coding tool.
Open-Source Community — led 10+ contributors through structured releases. 500+ monthly active users on kalaamlang.in.
TEDx Bangalore — invited to speak on "Why we should be able to code in our own languages." Recognised as a meaningful intervention in CS access for non-English speakers.

5-Phase Interpreter Pipeline — Pure JS · Zero Dependencies

Source

हिंदी / मराठी
বাংলা / తెలుగు...

→

① Cleaning

keyword map
→ normalized tokens

→

② Scanning

char-level scan
→ cleaned_sourcedata[]

→

③ Tokenization

20 Push* functions
→ typed tokens[]

→

④ Interpretation

walk tokens[]
→ memory{} + ExecutionStack[]

→

⑤ Output

kalaam{}: output
trace · timing

Learning Mode: ExecutionStack[] replays every step in the student's language — the trace is the product. New language = 1 keyword map entry, zero parser changes. npm package `kalaam` v2.3.3.

Custom Lexer Interpreter Vue 2 Quasar PWA npm

System Design

P.07

stringy-core — JavaScript String Utility Library

npm · 50+ pure functions · 9 modules · 19 forks · zero dependencies

ESM with named tree-shakeable exports and a unified _s namespace. Open-source contribution platform — stub functions invite community PRs.
19 forks. Published on npm.

9-Module Architecture — 50+ Functions · Zero Runtime Dependencies

textCaseManipulation

camelCase · snakeCase · titleCase

textCleaning

trim · normalizeWhitespace

textFormatting

Intl API — date · currency · relative

textMaskingAndSecurity

maskEmail · maskPhone · moderate

textMetadataAndExtraction

URLs · IPs · emails · hashtags

textAnalysisAndValidation

isPalindrome · matchesPattern

textTransformations

shorten · wordWrap · shuffle

textSpecializedOperations

levenshteinDistance · isBalanced

textGeneration

randomString · loremIpsum

Two import patterns: import { maskEmail } from 'stringy-core' (tree-shakeable) · import { _s } (namespace convenience). Husky + lint-staged pre-commit.

ESM Zero deps Jest 9 modules

System Design

P.08

Bharatvarsh.art — D2C E-Commerce Platform

Indian cultural wall art · Built and led engineering end-to-end

Architecture, payments, inventory, and delivery integrations from scratch.
₹2Cr+ in revenue.

E-commerce Payments Full-stack

Systems that scale.
AI that ships.
Code that matters.

Eight projects. Real outcomes.

Let's build
something.

Systems that scale. AI that ships. Code that matters.

Eight projects. Real outcomes.

Let's buildsomething.

Systems that scale.
AI that ships.
Code that matters.

Let's build
something.