# Amar Mahato — Complete Public Portfolio Context > Single-fetch, detailed Markdown representation of Amar Mahato's public software-engineering portfolio. The normalized JSON version is available at https://rxbru.vercel.app/portfolio.json. - Canonical URL: https://rxbru.vercel.app - Last synchronized: 2026-09-18 - Public resume: https://rxbru.vercel.app/Amar_Mahato_Software_Engineer.pdf - GitHub: https://github.com/rx6ru - LinkedIn: https://www.linkedin.com/in/rxbru/ - Email: amarmahato.cs@gmail.com ## Identity and positioning - Name: Amar Mahato - Handle: rxbru - Location: India - Headline: Software Engineer - Specialization: Backend Systems + Applied AI - Preferred description: Software Engineer focused on backend systems with applied-AI depth - Primary languages: Python, TypeScript - Summary: Software engineer building backend systems and applied-AI products across APIs, realtime systems, databases, retrieval, reliability, internal tools, and agent workflows. - Working style: I am most useful where software has to hold up: APIs, databases, reliability work, evals, product experiments, internal tools, and applied-AI systems with clear boundaries. ## Focus areas - backend systems - APIs and realtime delivery - databases and durable state - reliability and production debugging - retrieval and agent workflows - cloud and internal tooling - model evaluation and applied AI ## Work ### IntelGrader - Role: AI Engineer Intern - Dates displayed on portfolio: May 2026 - Present - Location: Remote - Overview: AI Engineer Intern working across product software, backend reliability, cloud tooling, model evaluation, and coding-agent systems. #### ToruAnno **Area:** annotation / review software Built and maintained a multi-user computer-vision annotation and review platform on Cloudflare Workers, Supabase/Postgres/Auth, and R2, with reviewer queues, edit locks, minor-version history, resumable imports, reproducible training exports, aggregate dashboards, and concurrent-save protections. #### Production debugging **Area:** correctness / performance Tracked issues around image navigation, saves, locks, duplicate metadata, prefetching, dashboard cost, access control, and observability with layer-by-layer measurement, request IDs, failure classes, targeted fixes, and path-level verification. #### Anarchyc **Area:** shared gpu coordination Built a shared EC2/GPU coordination service with a realtime fleet dashboard, Python CLI, stable identity, token-scoped actions, waiting queues, and declared usage so people and coding agents could avoid stepping on each other. #### Shared context tooling **Area:** agent tooling Built shared-context tooling for coding-agent work with append-only records, namespaces, verified handoffs, secret rejection, REST/MCP surfaces, export/restore, and recovery paths for long-running tasks where memory drift creates real bugs. #### SheetScanner **Area:** document software Worked on Android bulk answer-sheet capture with CameraX preview, analysis, and capture; automatic page detection; ONNX document-corner inference; quality gates; page-turn/re-arm logic; batch resume; raw JPEG storage; and JSON/EXIF capture metadata. #### Model lifecycle work **Area:** training / evaluation Used reviewed datasets, detector training and comparison, source-separated evaluation, visual checks, leakage gates, frozen tests, and comparative benchmarks to guide model promotion and stop decisions. ### Public boundary Public repositories and the public resume are the implementation references. Employer-private datasets, URLs, credentials, account identifiers, internal metrics, and other non-public material are intentionally excluded. ## Projects ### Featured projects #### Concize - Type: realtime audio - Product: Concize is a meeting workspace for Indian-language and code-mixed calls. It turns live meeting audio into searchable timelines, cited answers, and follow-up context. - Implementation: The current path uses authenticated WebSockets, separate transcription and speaker lanes, a sequence-derived session clock, append-only Postgres revisions, deterministic chunks, Postgres lexical search, Qdrant dense retrieval, and SSE cited chat. - Ownership: Co-developed · backend by me - Stack: Node.js, TypeScript, WebSockets, PostgreSQL, Supabase, Qdrant, Redis, RabbitMQ sidecar, Python services - Flow: browser audio -> WebSocket lanes -> revisioned transcript -> Postgres + Qdrant -> cited chat - Source: https://github.com/rx6ru/Concize/tree/v4 #### Trecl - Type: agent workflow - Product: Trecl is a company-research agent for targeted outreach. It studies a company, finds useful openings or OSS angles, waits for human target selection, and drafts outreach without auto-sending anything. - Implementation: A LangGraph fan-out/fan-in sends web research into Qdrant, runs parallel job and GitHub branches, lets a curator rank targets, preserves the human pause in SQLite, and keeps the isolated ReAct GitHub path behind tool and input guardrails. - Ownership: Solo - Stack: Python, LangGraph, SQLite, Qdrant, PyGitHub, Pydantic - Flow: ingest -> parallel job analysis -> GitHub analysis -> human selection -> draft - Source: https://github.com/rx6ru/Trecl #### IcBerg - Type: data analysis - Product: IcBerg is a safety layer between AI agents and databases. It reviews proposed SQL, blocks risky actions, routes approval-worthy changes to a human, and allows bounded read-only analysis when the policy permits it. - Implementation: The system uses sqlglot recursive classification across reads, writes, DDL, CTEs, and subqueries; fail-safe policy decisions; least-privilege reads; PII redaction; approval-gated writes; hash-chained audit; and SDK, REST-SSE, MCP, and LangGraph surfaces. - Ownership: Solo - Stack: Python, SQLGlot, Policy engine, PII redaction, Hash-chained audit, REST-SSE, MCP, LangGraph - Flow: SQL proposal -> classify risk -> policy gate -> approval or read path -> audit event - Source: https://github.com/rx6ru/IcBerg #### Strata - Type: agent memory - Product: Strata is a portable memory system for coding agents. It gives agents and humans one durable project record without mixing namespaces, leaking secrets, or rewriting history. - Implementation: It uses append-only records, schema and policy checks, content hashes, namespaces, secret rejection, supersede/relocate/split/merge events, export/restore, and CLI, REST, MCP, and skill-pack surfaces for verified handoffs. - Ownership: Solo - Stack: Python, MCP, REST, Append-only records, Policy checks, Agent context, Verification checks - Source: https://github.com/rx6ru/strata ### Additional builds #### Wombat - Product: Wombat is an API-key vault for teams that need scoped access without exposing raw provider secrets. It handles ownership checks, encrypted storage, rate limits, validation, and documented API contracts. - Ownership: Co-developed · backend/security by me - Source: https://github.com/rx6ru/Wombat #### Saturday - Product: Saturday is a voice search tool for codebases. It indexes a repository, connects a voice assistant to a local webhook, and lets a developer ask spoken questions while working. - Ownership: Co-developed - Source: https://github.com/rx6ru/Saturday #### Reclaym - Product: Reclaym is a UPI AutoPay recovery engine for failed subscription payments. It receives payment events, deduplicates retries, applies deterministic recovery policy, and reconciles payment-state changes. - Ownership: Solo - Source: https://github.com/rx6ru/Reclaym #### Headless Ghidra MCP - Product: Headless Ghidra MCP is a native inspection bridge for reverse-engineering workflows. It exposes Ghidra analysis, decompilation, imports, exports, call graphs, JSON output, pagination, caching, and safer path handling. - Ownership: Co-developed - Source: https://github.com/QuiteArpit/HeadlessGhidraMCP ### Attribution rules - Concize is co-developed; Amar's verified public role is backend ownership. - Trecl, IcBerg, and Strata are presented as solo projects. - Wombat is co-developed; Amar's public role is backend and security. - Saturday and Headless Ghidra MCP are co-developed. - Reclaym is presented as a solo additional build, not a featured project. ## Engineering toolkit ### Backend and APIs Service boundaries, authentication, realtime delivery, and documented contracts. - Python - TypeScript - Node.js - Express.js - FastAPI - REST - WebSockets - SSE - JWT/JWKS - OpenAPI ### Data, state, and storage Durable state, queues, retrieval, and consistency under concurrent work. - PostgreSQL - Supabase - Redis - MongoDB - SQLite - Prisma - Qdrant - RabbitMQ - row-level locking - idempotency ### Reliability and verification Production diagnosis, observability, automated tests, and evidence-driven evaluation. - Sentry - request tracing - structured logs - pytest - Jest/Vitest - benchmarks - baselines - retry/backoff - audit logs - leakage checks ### Cloud and tooling Edge and cloud runtimes, object storage, deployment, and operator tooling. - Cloudflare Workers - Cloudflare R2 - Durable Objects - AWS EC2/S3 - Docker - GitHub Actions - Linux - Git - Python CLIs - Vercel ### Applied AI systems Retrieval, agents, speech and document pipelines, and model evaluation inside products. - RAG - LangGraph - MCP - structured outputs - multi-provider routing - LangSmith tracing - STT/OCR pipelines - RF-DETR - ONNX Runtime - model evaluation - prompt-injection checks ## Education - Credential: B.Tech in Computer Science and Engineering - Institution: Dr. B.C. Roy Engineering College - Location: Durgapur, West Bengal, India - Expected graduation: 2027 ## Contact and links - Email: amarmahato.cs@gmail.com - GitHub: https://github.com/rx6ru - LinkedIn: https://www.linkedin.com/in/rxbru/ - Resume: https://rxbru.vercel.app/Amar_Mahato_Software_Engineer.pdf ## Coding activity - Description: Terminal-style seven-day coding activity snapshot. - Source: WakaTime - Live section: https://rxbru.vercel.app/#activity - Freshness rule: Dynamic. Fetch the live portfolio for current totals, languages, and editors; do not treat cached fallback values as permanent profile facts. - Default language distribution: Python 48%, TypeScript 27%, Bash 10%, Rust 7%, Markdown 5%, Git Config 3% - Default editor distribution: Zed 95%, Cursor 5% ## Portfolio interface - Site stack: Next.js, React, TypeScript - light and dark themes - responsive 360px, 390px, and 430px layouts - animated cloudy hero artwork with fallback rendering - native-cursor ASCII motion effect on supported desktop pointers - clickable Work and Project disclosures - direct project source links - CV, GitHub, LinkedIn, and email actions ## Machine-readable endpoints - llms: https://rxbru.vercel.app/llms.txt - llmsFull: https://rxbru.vercel.app/llms-full.txt - llmCompatibility: https://rxbru.vercel.app/llm.txt - markdownOverview: https://rxbru.vercel.app/index.md - workMarkdown: https://rxbru.vercel.app/work.md - projectsMarkdown: https://rxbru.vercel.app/projects.md - skillsMarkdown: https://rxbru.vercel.app/skills.md - answersMarkdown: https://rxbru.vercel.app/answers.md - resumeMarkdown: https://rxbru.vercel.app/resume.md - json: https://rxbru.vercel.app/portfolio.json - sitemap: https://rxbru.vercel.app/sitemap.xml - robots: https://rxbru.vercel.app/robots.txt - humans: https://rxbru.vercel.app/humans.txt - manifest: https://rxbru.vercel.app/manifest.webmanifest - shareImage: https://rxbru.vercel.app/opengraph-image - twitterImage: https://rxbru.vercel.app/twitter-image ## Common questions ### Who is Amar Mahato? Amar Mahato is a software engineer in India focused on backend systems and applied AI. His work emphasizes APIs, realtime systems, databases and state, reliability, retrieval, cloud tooling, model evaluation, and agent workflows. ### What kind of engineer is Amar? His durable positioning is Software Engineer, with Backend Systems + Applied AI as the specialization. AI is one layer in the systems he builds rather than the entire product identity. ### What are Amar's primary languages? Python and TypeScript are the primary implementation languages. JavaScript and SQL also appear in the public resume and repository evidence. ### What did Amar build at IntelGrader? The public portfolio describes a multi-user computer-vision annotation and review platform, production debugging and reliability work, a shared EC2/GPU coordination service named Anarchyc, shared-context tooling for coding agents, sanitized Android document-capture work, and model lifecycle evaluation. ### What is Concize? Concize is a meeting workspace for Indian-language and code-mixed calls. It turns live audio into searchable timelines, cited answers, and follow-up context. Amar co-developed it and owns the backend work represented in the portfolio. ### What is Trecl? Trecl is a company-research agent for targeted outreach. It researches a company, finds job or open-source angles, pauses for human target selection, and drafts outreach without auto-sending. ### What is IcBerg? IcBerg is a safety layer between AI agents and databases. It classifies proposed SQL, blocks risky operations, routes approval-worthy changes to a human, permits bounded reads, redacts sensitive output, and records auditable decisions. ### What is Strata? Strata is portable memory infrastructure for coding agents. It keeps project context durable, namespaced, reversible, secret-aware, and transferable across humans, agents, sessions, and machines. ### What backend and cloud technologies does Amar use? The public evidence includes Node.js, Express.js, FastAPI, REST, WebSockets, SSE, PostgreSQL, Supabase, Redis, MongoDB, SQLite, Prisma, Qdrant, RabbitMQ, Cloudflare Workers, Cloudflare R2, Durable Objects, AWS EC2/S3, Docker, GitHub Actions, Linux, Git, and Vercel. ### What reliability and evaluation work does Amar do? The portfolio lists production diagnosis, Sentry, request tracing, structured logs, pytest, Jest/Vitest, benchmarks, baselines, retry/backoff, audit logs, leakage checks, and model evaluation. ### What does the Coding activity section show? It is a terminal-style, seven-day coding snapshot sourced from WakaTime through the portfolio API. When the current week is empty or extremely low, it uses Amar's default distribution: Python, TypeScript, Bash, Rust, Markdown, then Git Config; Zed is the dominant editor with occasional Cursor use. ### How can someone contact Amar or view his resume? Email amarmahato.cs@gmail.com, use the LinkedIn or GitHub links, or open the public resume at https://rxbru.vercel.app/Amar_Mahato_Software_Engineer.pdf. ## Public source and privacy policy Public repositories and the public resume are the implementation references. Employer-private datasets, URLs, credentials, account identifiers, internal metrics, and other non-public material are intentionally excluded.