---
title: "Aman Arun - Independent Project in Bengaluru | Tal"
description: "Meet Aman Arun, Staff/Lead Backend Engineer | Distributed Systems at Scale | Java, Spring Boot, Kafka, AWS | Agentic AI & Production LLM Systems | Ex-Medibuddy, Clear, Intuit | IIT Guwahati based in Bengaluru. AI stack includes antigravity, claude code, and langchain."
image: "https://tal.club/meet/aman-arun/share-image"
---

# Aman Arun on Tal

## Headline

Staff/Lead Backend Engineer \| Distributed Systems at Scale \| Java, Spring Boot, Kafka, AWS \| Agentic AI & Production LLM Systems \| Ex\-Medibuddy, Clear, Intuit \| IIT Guwahati

## About

Backend and platform engineer with 11 years of building distributed systems at scale, from pipelines reconciling 100M\+ records to event streams handling 200M daily events\. I've led 0\-to\-1 platform builds and teams of up to 8 as the sole technical owner, and shipped two production LLM systems that serve 25K\+ daily decisions\. What I keep coming back to is building the second and third use cases into the first\. Most of my work has turned one\-off solutions into platforms other teams build on — a provider\-agnostic integration SPI, a database\-configured rule engine, a reconciliation engine reused across matching use cases, an authorization library adopted as the org standard across 8 services\. Areas I work in: \- Distributed systems and high\-throughput data pipelines \(Kafka, Trino, Kinesis, Temporal\) \- Backend architecture and platform design \(Java, Spring Boot, Python, Node\.js, microservices\) \- Production AI systems \(LLM pipelines, RAG, agentic workflows, evals, MCP\) \- Search and query infrastructure \(OpenSearch, ClickHouse, PostgreSQL, MongoDB\) \- Technical leadership: architecture ownership, cross\-team standards, mentoring Currently building gotrades\.in, a live intraday algorithmic trading system on a two\-service architecture handling 100 concurrent instrument tick streams with an ML inference service backing execution\. B\.Tech from IIT Guwahati\. Open to Staff and Lead backend engineering roles\.

## Location

Bengaluru

## AI stack

antigravity

claude code

langchain

langgraph

mcp

## Work experience

Independent Project at Gotrades\.in, Jan 2026 \- Present · 8 mos, Bengaluru, Built and operated a live intraday algorithmic trading system on a two\-service architecture\. 🔹 Trader Service: Live execution engine subscribing to 100 concurrent instrument tick streams at millisecond update granularity, evaluating each candle for entry signals and managing the full order lifecycle through fulfilment and exit, with a hard flat\-by\-close invariant on all intraday positions\. Running unattended daily since January 2026\. 🔹 Trainer Service: Machine learning service training and serving a neural network over engineered price, volume, and technical\-indicator features, deliberately optimised for precision over recall to minimise false entry signals, with normalisation designed to prevent forward\-looking data leakage and real\-time per\-candle inference served to the execution engine\.

Tech Lead at MediBuddy, Mar 2024 \- Dec 2025 · 1 yr 10 mos, Bengaluru, Led technical direction for the in\-clinic vertical and platform work across Consults, Labs, and Meds\. 🔹 InClinic Vertical: Owned technical direction for a 0\-to\-1 Inclinic consultation business as sole tech lead of a team of 6, scaling 250 to 3,000\+ daily orders \(5,500 peak\) and 8K to 50K doctors\. Developed a better discovery layer for fuzzy matching, filtering, and sorting\. 🔹 Provider Integration Platform: Provider\-agnostic booking layer \(strategy\-pattern SPI\) serving real\-time availability for 1,200\+ doctors across 4 providers, cutting onboarding from 2\+ weeks to 2\-3 days with zero\-downtime Apollo/Medicover launches\. Insulated the funnel from unreliable vendor APIs via a 60s TTL cache, circuit breakers, and backoff retries; deferred confirmation eliminated double\-booking on timeouts\. Drove 50% order conversion increase\. 🔹 Prescription Validation Agent: Async LLM pipeline validating 25K daily prescriptions against order context, auto\-clearing 76% and cutting 60% of manual validation cost\. An asymmetric confidence gate biased against auto\-approval escalated uncertain cases to human ops, with multi\-model fallback, retries, and a continuous eval harness shadow\-auditing 10% of decisions at 83% accuracy\. 🔹 Search Infrastructure: Fixed chronic OpenSearch instability that restarted nodes under peak load, taking down search across all three verticals\. Traced JVM heap pressure to fielddata on text fields, idle indices, and over\-sharding, validating fixes through a shadow cluster mirroring production traffic and load tests at 6\.7x throughput\. Halved response times and cut peak\-month cost \~50%, scaling data nodes 8 to 4\. 🔹 Provider Central: Multi\-tenant provider control plane moving 1,200\+ daily bookings off a phone workflow, eliminating \~1,000 manual confirmation calls/day for 2,000\+ operators\. Hierarchical org\-branch\-operator permission model with specialization\-scoped access\.

Lead Software Engineer at ClearTax, Jun 2022 \- Mar 2024 · 1 yr 10 mos, Bengaluru, 🔹 File Ingestion Platform: Streaming CSV pipeline processing 1M row files in 3\-5 minutes \(\~5K rows/sec\), scaling to 5M row uploads from enterprise clients like Zepto and Trend\. 100\+ MVEL validation and tax rules configured in MongoDB shipped business logic as config rather than releases\. Returned per\-row error reasons while still processing valid rows, and isolated large and small tenants into separate processing channels with checksum\-based dedupe to prevent head\-of\-line blocking across 250\+ clients\. 🔹 Analytics GPT: Agentic text\-to\-SQL system used daily by 50\+ PMs and client\-facing analysts, cutting insight turnaround from 48 hours to under a minute\. Grounded generation through a RAG layer over vector\-indexed schema documentation retrieved per query, with an agent that routed between PostgreSQL and MongoDB, selected the execution tool, self\-corrected on failed queries, and enforced mandatory org\-ID scoping to keep results bounded to authorized client accounts\. 🔹 EwayBill Reconciliation: Trino\-based reconciliation engine matching 100M\+ e\-way bills against sales documents in 8\-10 minutes, querying S3 directly to skip warehouse loading and stay cloud\-agnostic\. Generalized into a reusable matching platform powering multiple new reconciliation use cases without engine changes, with retry and cluster\-recycle recovery for memory exhaustion at peak\. 🔹 Clear Assurance: Compliance analytics product built from scratch for 20\+ enterprise Tax Heads on a ClickHouse store with embedded Superset, using Temporal workflows for durable multi\-stage ingestion \(source DB to Parquet to ClickHouse\) that resumed from failure mid\-pipeline\.

Senior Software Engineer at Intuit India, May 2021 \- Jun 2022 · 1 yr 2 mos, Bengaluru, 🔹 Notification Platform: Eliminated broken object\-level authorization across 8 microservices with a centralized AOP\-based authorization library that reduced integration to a single annotation, adopted as the org's mandatory standard for all new services\. 🔹 Communication Hub: Built an event ingestion pipeline streaming 200M peak daily events through AWS Kinesis into a Hadoop data lake, with at\-least\-once delivery and notification\-ID deduplication downstream to keep analytics consumers idempotent\. 🔹 Widget Integration: Built a multi\-tenant notification UI engine on a micro\-frontend architecture, shipped as a plug\-and\-play module integrated across 20 products in the Intuit ecosystem\.

Senior Software Engineer at Medlife\.com, Sep 2018 \- Apr 2021 · 2 yrs 8 mos, Bengaluru, 🔹 Payments Platform: Led a 4\-engineer team building a gateway\-agnostic payment layer over Juspay, Razorpay, and Paytm for 25K\+ daily orders, auto\-switching gateways on elevated failure rates to cut transaction errors from 12% to 1%, with order\-ID idempotency and scheduled reconciliation preventing double charges\. 🔹 Customer App Migration: Started and prototyped a ground\-up React Native rewrite of the legacy Cordova app, then grew and led a team of 8 to ship it with expanded features and no disruption to existing user flows, driving a 20% increase in customer orders\. 🔹 Purchase Order & Procurement: Built an event\-driven auto\-procurement system with Just\-In\-Time ordering integrations and automated CSV/email processing, plus a preventive\-alert tool that saved millions in monthly dead stock\.

Co\-Founder at Naradd, Jul 2016 \- Sep 2018 · 2 y 2 mo, Bengaluru

Software Engineer at Virtusa, Aug 2015 \- Jun 2016 · 10 mo, Bengaluru

## Education

Bachelor of Technology \(B\.Tech\.\), Civil Engineering, Indian Institute of Technology, Guwahati

[View the original profile](https://tal.club/meet/aman-arun)

```json
{
  "@context": "https://schema.org",
  "@graph": [
    {
      "@type": "Person",
      "@id": "https://tal.club/meet/aman-arun#person",
      "name": "Aman Arun",
      "url": "https://tal.club/meet/aman-arun",
      "image": "https://tal.club/meet/aman-arun/profile-photo.jpg",
      "description": "Meet Aman Arun, Staff/Lead Backend Engineer | Distributed Systems at Scale | Java, Spring Boot, Kafka, AWS | Agentic AI & Production LLM Systems | Ex-Medibuddy, Clear, Intuit | IIT Guwahati based in Bengaluru. AI stack includes antigravity, claude code, and langchain.",
      "jobTitle": "Independent Project",
      "worksFor": {
        "@type": "Organization",
        "name": "Gotrades.in"
      },
      "hasOccupation": {
        "@type": "Occupation",
        "name": "Independent Project"
      },
      "homeLocation": {
        "@type": "Place",
        "name": "Bengaluru"
      },
      "alumniOf": [
        {
          "@type": "EducationalOrganization",
          "name": "Indian Institute of Technology, Guwahati"
        }
      ],
      "knowsAbout": [
        "antigravity",
        "claude code",
        "langchain",
        "langgraph",
        "mcp"
      ],
      "mainEntityOfPage": {
        "@id": "https://tal.club/meet/aman-arun#webpage"
      }
    },
    {
      "@type": "ProfilePage",
      "@id": "https://tal.club/meet/aman-arun#webpage",
      "url": "https://tal.club/meet/aman-arun",
      "name": "Aman Arun on Tal",
      "description": "Meet Aman Arun, Staff/Lead Backend Engineer | Distributed Systems at Scale | Java, Spring Boot, Kafka, AWS | Agentic AI & Production LLM Systems | Ex-Medibuddy, Clear, Intuit | IIT Guwahati based in Bengaluru. AI stack includes antigravity, claude code, and langchain.",
      "dateModified": "2026-08-24T16:50:47.815Z",
      "isPartOf": {
        "@id": "https://tal.club/#website"
      },
      "mainEntity": {
        "@id": "https://tal.club/meet/aman-arun#person"
      },
      "primaryImageOfPage": {
        "@type": "ImageObject",
        "url": "https://tal.club/meet/aman-arun/profile-photo.jpg",
        "contentUrl": "https://tal.club/meet/aman-arun/profile-photo.jpg",
        "caption": "Aman Arun profile photo on Tal"
      },
      "inLanguage": "en-IN"
    }
  ]
}
```