# Adhikram Maitra > Founding engineer at ANAX ($100K+ deal revenue, ~10x deal flow). Founder of TechPrufer (~2k uniques/day). Data platforms at Media.net (3B-row checks). Side products: Equal Collective ($20K+ MRR) and GripGrams. - Site: https://adhikram.com - Location: India — UTC+5:30 - Email: adhikrammaitra@gmail.com - Profiles: https://github.com/Adhikram, https://www.linkedin.com/in/adhikram-maitra/, https://x.com/Adhikram ## About Most mornings I am still on a New York broker's screen, working remote from India. At ANAX I am founding engineer on the commercial real-estate dashboard they close on. RAG over 20K+ financial documents, a chat they can stay in for the whole deal, outreach that lifted qualified deal flow about 10x, and more than $100K through a loop the desk can run after I leave the chair. The product I operate in public is TechPrufer. Job intel plus a tailoring workbench for people who already live in an editor. 400+ company trails, an MCP that Cursor and Claude can call, Chrome capture, LaTeX resumes. About 2,000 unique people use it on a given day. Two other products sit outside that job. GripGrams is in my hands now. A free nutrition and strength app for a household, Expo on both phones, no paywall. Equal Collective was the seller-ops SaaS I took from a blank repo to $20K+ MRR, with 30K+ listings enriched so outreach had something true to send. I learned to trust a pipeline the long way. Four years on Media.net's ad-tech data plane at Directi. Kafka at 50k records a second, Spark jobs that compared nearly 3 billion rows in 30 minutes, streaming audits that cut latency about in half. Before that, Devsnest. A community backend that reached 10K users in three months while notifications held at 2K messages a day. I want the next seat that still starts in someone else's workflow. Sit with the operator, ship agents and RAG against live systems, leave a number they can keep moving. Looking forward to forward-deployment opportunities. Desks first, then agents, then a number someone else can still move. ## How I work - 01 Sit in the seat: Brokers, sellers, job seekers. I work from their screen, not a backlog they never open. - 02 Name the number: Deal flow, latency, MRR, applications sent. If it does not move, it is not the job. - 03 Ship on their stack: Next.js, Python, MCP, Spark: whatever they already run. Agents and RAG only where they pay rent. - 04 Leave the loop: MCP, evals, docs. The next person should ship faster than I did on day one. ## Experience ### ANAX — Founding engineer Apr 2025 - present · https://anaxrep.com/ Commercial real-estate dashboard in New York. I sit with brokers and ship against live desk data: RAG, Broker Buddy chat, AI outreach, and a velocity engine they run the same week. - 20K+ financial docs in RAG - ~10x qualified deal flow / month - $100K+ deal revenue through the loop - Shipped agentic layers against live broker data and undocumented internal REST APIs, not a sandbox. RAG over 20K+ financial docs so Gemini-backed agents answer from real files. - Broker Buddy: a chat assistant for the full deal lifecycle. Brokers stay in conversation instead of clicking through the product. I tuned a stable version with evals and a harness. - AI outreach funnel lifted qualified deal flow ~10x by iterating targeting with brokers. Lead generation scaled 50K → 800K with Playwright, Gemini, and RAG on messy newsletter data. - Broker Velocity Engine plus ETL for 10+ brokers on 1,000+ deals and $100K+ revenue, so the desk still runs after handoff. Stack: Next.js, TypeScript, PostgreSQL, RAG, Python, Redis ### Media.net — Data Application Developer 2 Oct 2021 - Mar 2025 · https://www.media.net/ Ad-tech data plane at Directi. Streaming, batch, and data quality on billions of rows: DANTE for integrity, StopLoss for real-time audit. This is why I will not drown in an enterprise warehouse. - 50k/s Kafka ingest - ~50% streaming latency cut - ~3B rows compared in 30 minutes - DANTE, our first configurable data-quality suite: 60% faster on >100GB reads. EQ, Ratio, and PM tracers for metric trends, outliers, and mismatch coverage. - StopLoss: Spark Streaming to Structured Streaming, exactly-once semantics, ~50% latency down. A/B tests cut overall loss ~15%. - Compared nearly 3 billion and 200 million rows in 30 minutes. Kafka ingest at 50k records/s, processed in 5-minute buckets. Stack: Scala, Spark, Kafka, Hive, Java ### Devsnest — Backend engineer Oct 2020 - Oct 2021 · https://devsnest.in Ed-tech community backend. First founding-adjacent backend seat: Discord automation and a notification path that held as the product hit 10K users in three months. - 10K users in 3 months - 2K/day notification messages - Discord group-moderation automation through the web app so peer learning could run with less ops overhead. - Notification system at 2K messages/day as the community scaled to 10K users in three months. Stack: Ruby on Rails, Python, Redis, SQS ## Products Owned products, most recent first: GripGrams (currently shipping), TechPrufer, Equal Collective. Detail pages live at /products/. ### GripGrams — Founder 2025 - present · https://gripgrams.com/ Page: https://adhikram.com/products/gripgrams Currently shipping. Nutrition and strength-training app: kitchen, barcode, gym, family tracker. Expo / React Native with Firebase. Fully free. GripGrams is the product I am shipping now. It is a nutrition and strength-training app for a household, not a single gym log: kitchen, barcode, workouts, and a family tracker on the same graph. The app is Expo / React Native with a TypeScript monorepo. Auth and live backend sit on Firebase, including Cloud Functions. It is free at launch. No paywall. I built it as a complete mobile product, not a demo: meals, gym sessions, and household tracking that have to work on both iOS and Android through EAS. The constraint is the same as my other products: operators in a real kitchen or gym, not a slide about health tech. - iOS + Android Expo / EAS - Free no paywall at launch - Shipped a real mobile product: meals, workouts, and household tracking on a TypeScript monorepo. - Firebase auth and Cloud Functions for the live backend. Stack: Expo, React Native, Firebase, TypeScript ### TechPrufer — Founder Jan 2026 - present · https://techprufer.com/ Page: https://adhikram.com/products/techprufer Feature trail of TechPrufer surfaces (copy from the live product pages): company feature trails, Resume Editor, Job Tailor, Application Tracker, LinkedIn Improver, Chrome Extension. Local pages at /products/techprufer/. Public job intel plus a tailoring workbench: company blogs, interview trails, MCP across IDEs, Chrome capture, LaTeX resumes. I run this as a live GenAI product. TechPrufer is the public product I own. It is job intel plus a tailoring workbench: company blogs, interview trails, an MCP that other IDEs can call, Chrome capture, and a LaTeX resume engine. The data side is a scraper and clustering pipeline over open interview and blog sources. That bank now covers 400+ companies. The product side is a live site at about 2,000 unique users a day. The MCP server is the piece other agents use. Cursor, Claude, Codex, and Antigravity can tailor a profile against a JD without leaving the editor. The Chrome extension captures jobs, helps autofill, and renders a pixel-perfect LaTeX PDF. I ran the same loop for 150+ role-specific resumes. Stack is Next.js, TypeScript, Drizzle, PostgreSQL, Python, and MCP. I operate it as a GenAI product with a real user loop, not a portfolio toy. - ~2,000 unique users / day - 400+ companies in the bank - 150+ resumes through my pipeline - Scraper and clustering pipeline that turns open interview and blog sources into trails for 400+ companies. - MCP server so Cursor, Claude, Codex, and Antigravity can tailor a profile against any JD. - Chrome extension for job capture, autofill, and LinkedIn, plus a LaTeX PDF engine. I ran 150+ role-specific resumes through the same loop. Stack: Next.js, Drizzle, PostgreSQL, MCP, Python, TypeScript ### Equal Collective — Founding engineer Jan 2024 - Dec 2025 · https://www.linkedin.com/company/equal-collective/ Page: https://adhikram.com/products/equal-collective Seller-ops SaaS. I took AI outreach, listing enrichment, and analytics from zero to paid, self-sustaining revenue. Equal Collective is seller-ops SaaS. I was founding engineer from January 2024 through December 2025. The job was to take AI outreach, listing enrichment, and analytics from zero to paid revenue. I led it from inception to a self-sustaining state. Outreach and Metabase loops carried the product to $20K+ monthly revenue. Sellers paid because listings moved, not because a dashboard looked busy. On the data side I ingested Amazon reviews and listing data at scale, past the usual scrape defenses, then pushed enrichment back into seller reach. That covered 30K+ products and 30K+ sellers. Playwright handled capture. Python sat under the pipelines. Smartlead ran the send path. - 0 → $20K+ MRR - 30K+ sellers / listings enriched - Led the product from inception to a self-sustaining loop: AI outreach and Metabase carried it to $20K+ monthly revenue. - Advanced outreach funnel with AI-powered send and integrated analytics so the next campaign was better than the last. - Amazon ingestion at scale: reviews and listing data for 30K+ products, then enrichment back into seller reach for 30K+ sellers. Stack: Python, Smartlead, Metabase, Playwright ## Reading (TechPrufer) Attached from the live TechPrufer account. Feature trails I follow, then system designs I mark solved or save. Source: https://techprufer.com. ### Feature trails - [Cloudflare — Artificial Intelligence Integration & Impact](https://techprufer.com/blogs/cloudflare/artificial-intelligence-integration-impact) ### Designs I worked through - [Design a RAG system with evaluation](https://techprufer.com/companies/openai/design-a-rag-system-with-evaluation) — OpenAI, ML System Design - [Design and optimize a RAG system](https://techprufer.com/companies/openai/design-and-optimize-a-rag-system) — OpenAI, ML System Design - [Design a Distributed Rate Limiter](https://techprufer.com/companies/openai/design-a-distributed-rate-limiter-2) — OpenAI, System Design - [Design a sandboxed cloud IDE](https://techprufer.com/companies/openai/design-a-sandboxed-cloud-ide) — OpenAI, System Design - [How to stream a large file to 1000 hosts fastest](https://techprufer.com/companies/anthropic/how-to-stream-a-large-file-to-1000-hosts-fastest) — Anthropic, System Design - [Design a distributed web crawler](https://techprufer.com/companies/anthropic/design-a-distributed-web-crawler-2) — Anthropic, System Design - [Design a concurrent web crawler](https://techprufer.com/companies/anthropic/design-a-concurrent-web-crawler-2) — Anthropic, System Design - [Design an Online Coding Judge Platform](https://techprufer.com/companies/google/design-an-online-coding-judge-platform) — Google, System Design - [Google Calendar](https://techprufer.com/companies/google/lc-google-sd-google-calendar) — Google, System Design - [Concurrent real-time notification system](https://techprufer.com/companies/google/lc-google-sd-concurrent-real-time-notification-system) — Google, System Design - [Design Scalable Notification Rate Limiter](https://techprufer.com/companies/cursor/design-scalable-notification-rate-limiter) — Cursor, System Design - [Design a scalable tagging system](https://techprufer.com/companies/atlassian/design-a-scalable-tagging-system) — Atlassian, System Design - [Build and design a Mistral RAG agent](https://techprufer.com/companies/mistral-ai/build-and-design-a-mistral-rag-agent) — Mistral AI, ML System Design ### Also saved - [Message Logger](https://techprufer.com/companies/google/lc-google-sd-message-logger) — Google - [Detect Cycle in Spreadsheet](https://techprufer.com/companies/google/lc-google-sd-detect-cycle-in-spreadsheet) — Google - [Design an Ad Serving System](https://techprufer.com/companies/google/lc-google-sd-design-an-ad-serving-system) — Google - [Design Calendar Event Conflict Handling](https://techprufer.com/companies/google/design-calendar-event-conflict-handling) — Google - [Design a global real-time notification system](https://techprufer.com/companies/google/design-a-global-real-time-notification-system) — Google - [Design a Twitter hashtag metrics aggregator](https://techprufer.com/companies/google/design-a-twitter-hashtag-metrics-aggregator) — Google - [Design deduplicated file storage on filesystem](https://techprufer.com/companies/google/design-deduplicated-file-storage-on-filesystem) — Google - [Design a URL shortener for 100M users](https://techprufer.com/companies/google/lc-google-sd-design-a-url-shortener-for-100m-users) — Google - [Design Twitter](https://techprufer.com/companies/google/lc-google-sd-design-twitter) — Google - [Design Facebook](https://techprufer.com/companies/google/lc-google-sd-design-facebook) — Google - [Google Photos](https://techprufer.com/companies/google/lc-google-sd-google-photos) — Google - [Design a ride-sharing application](https://techprufer.com/companies/google/lc-google-sd-design-a-ride-sharing-application) — Google - [Design a payment system with holds and batching](https://techprufer.com/companies/openai/design-a-payment-system-with-holds-and-batching) — OpenAI - [Design a scalable payment processor](https://techprufer.com/companies/openai/design-a-scalable-payment-processor) — OpenAI - [Design a Multi-Tenant Online IDE](https://techprufer.com/companies/openai/design-a-multi-tenant-online-ide) — OpenAI - [Design a distributed crossword solving service](https://techprufer.com/companies/openai/design-a-distributed-crossword-solving-service) — OpenAI - [Design a crossword puzzle solver system](https://techprufer.com/companies/openai/design-a-crossword-puzzle-solver-system) — OpenAI ## Open-source forks Public GitHub forks of open-source software I used: Firecrawl, Crawl4AI, and LinkedIn Scraper. - [firecrawl-cre](https://github.com/Adhikram/firecrawl-cre) — Firecrawl fork. Web data API, adapted for CRE scrape and extract. Upstream: firecrawl/firecrawl (https://github.com/firecrawl/firecrawl). - [crawl4ai-cre](https://github.com/Adhikram/crawl4ai-cre) — Crawl4AI fork. LLM-ready crawler with CRE crawler enhancements. Upstream: unclecode/crawl4ai (https://github.com/unclecode/crawl4ai). - [linkedin_scraper](https://github.com/Adhikram/linkedin_scraper) — LinkedIn scraper fork. Async Playwright for profiles, companies, and jobs. Upstream: joeyism/linkedin_scraper (https://github.com/joeyism/linkedin_scraper). ## Stack ### AI & agents Agentic AI, Agent Orchestration, RAG, MCP, Gemini, Harness, Agent Eval, Multi-LLM Integration, LLM Fine-Tuning, Vector Databases, Cursor, Claude Code ### Languages TypeScript, Python, JavaScript, SQL, Scala, Java, Ruby, C++ ### Backend & frameworks Next.js, Node.js, PostgreSQL, Drizzle, Express, REST APIs, Ruby on Rails, Microservices, Astro ### Cloud & DevOps AWS, Docker, Redis, SQS, Git, Nginx, Jenkins, EC2, EMR, Azure ### Data engineering Spark, Kafka, Hive, Structured Streaming, PySpark, HDFS, ETL, Data Modeling, MySQL ### Analytics Metabase, A/B Testing, Data Pipelines, Real-time Analytics, Performance Monitoring ### Systems System Architecture, Distributed Systems, Design Patterns, OOP, Algorithms ### Mobile Expo, React Native, Firebase, Linux, pm2 / VPS ## Areas of focus Sit with the operator, Move a number, not a demo, Agentic RAG in the product, MCP in the editor, Data platforms at billions of rows, Live GenAI with a user loop, Mobile that actually ships ## Common questions ### Who is Adhikram Maitra? Most mornings I am still on a New York broker's screen, working remote from India. At ANAX I am founding engineer on the commercial real-estate dashboard they close on. RAG over 20K+ financial documents, a chat they can stay in for the whole deal, outreach that lifted qualified deal flow about 10x, and more than $100K through a loop the desk can run after I leave the chair. The product I operate in public is TechPrufer. Job intel plus a tailoring workbench for people who already live in an editor. 400+ company trails, an MCP that Cursor and Claude can call, Chrome capture, LaTeX resumes. About 2,000 unique people use it on a given day. Two other products sit outside that job. GripGrams is in my hands now. A free nutrition and strength app for a household, Expo on both phones, no paywall. Equal Collective was the seller-ops SaaS I took from a blank repo to $20K+ MRR, with 30K+ listings enriched so outreach had something true to send. I learned to trust a pipeline the long way. Four years on Media.net's ad-tech data plane at Directi. Kafka at 50k records a second, Spark jobs that compared nearly 3 billion rows in 30 minutes, streaming audits that cut latency about in half. Before that, Devsnest. A community backend that reached 10K users in three months while notifications held at 2K messages a day. I want the next seat that still starts in someone else's workflow. Sit with the operator, ship agents and RAG against live systems, leave a number they can keep moving. Looking forward to forward-deployment opportunities. ### What is ANAX? Commercial real-estate dashboard in New York. I sit with brokers and ship against live desk data: RAG, Broker Buddy chat, AI outreach, and a velocity engine they run the same week. ### What is TechPrufer? Public job intel plus a tailoring workbench: company blogs, interview trails, MCP across IDEs, Chrome capture, LaTeX resumes. I run this as a live GenAI product. ### Is Adhikram available for work? Yes. Looking forward to forward-deployment opportunities. Fastest contact: adhikrammaitra@gmail.com. ### What timezone is Adhikram in? India — UTC+5:30