Satish

Full-Stack Developer

Mumbai, India 0+ yrs exp 82 · Excellent

About

Computer Engineering graduate (CGPA 9.81/10) with internships at Kodo and BrowserStack AI Labs, experienced in full-stack web development, LLM agents, Android app development, and automation tooling. Proficient in Python, React.js, Node.js, Express.js, and Kotlin with a strong foundation in DSA, DBMS, and Operating Systems. Passionate about building scalable, user-centric products.

Skills & Expertise (45)

Python Intermediate

7.5/10

Years Exp

JavaScript Intermediate

7.0/10

Years Exp

Typescript Intermediate

7.0/10

Years Exp

Node.js Intermediate

6.5/10

Years Exp

React.js Intermediate

6.5/10

Years Exp

Postgresql Intermediate

6.0/10

Years Exp

FastAPI Intermediate

6.0/10

Years Exp

Celery Intermediate

5.5/10

Years Exp

Git Intermediate

5.0/10

Years Exp

Data Structures & Algorithms Intermediate

5.0/10

Years Exp

Docker Intermediate

5.0/10

Years Exp

Puppeteer Selenium Playwright ClickHouse Firebase Redis RabbitMQ Nodemailer GitHub Azure Blob AWS CloudWatch Kibana DBMS Operating systems Compiler Design OOP Prompt Engineering Java kotlin LangGraph LangChain Claude OpenAI APIs AI Agents Tool Calling MongoDb Next.js Jetpack Compose Tailwind CSS Express.js REST APIs C++ Android Development MySql

Work Experience

Software Engineer Intern — Agentic AI Team

Kodo Technologies

Jul 2026 - Oct 2026

Shipped features for a production LLM treasury agent (Python, FastAPI, LangGraph, Claude) on web and WhatsApp. Launched the email channel for proactive nudges alongside WhatsApp (Celery on RabbitMQ), redesigning PostgreSQL tables to per-channel delivery rows so each channel retries independently; added per-company nudge thresholds. Rebuilt the agent’s orders tool on the read replica, scoped to the caller, surfacing 38% of orders previously hidden from the agent and fixing pending-approval counts overstated 4x+. Delivered chat-based order actions (withdraw, approve, reject) on web and WhatsApp behind a two-step, state-backed confirmation gate keyed by action and order, so one confirmation can never authorize a different action. Added agent tools that read the replica directly for fund details across 7,600+ schemes, bank FD rates and transaction history, and exposed 3 paginated REST APIs that serve bank-statement data to the web app. Fixed wrong figures in financial advice: excluded stale bank balances from spendable cash, capped reminder delivery to a 15-minute window after outages, and brought the agent’s fund returns to full parity with the app. Cut a transaction-history reply from 76s to 5s and prompt tokens 4x by removing redundant tool calls, verified with LLM evals. Debugged production nudge-delivery failures on WhatsApp and email, tracing Kibana logs and DB records to root causes.

SDE Intern — AI Labs Team

BrowserStack

Jan 2026 - Jul 2026

Engineered a granular Rerun feature enabling re-execution of specific items, evaluators, or item-evaluator pairs within experiments, eliminating full-experiment reruns and reducing user API token costs by up to 80%. Designed and shipped Session-Level Aggregation, grouping any number of traces under a single session ID into one session-wide view, delivering observability and metrics previously inaccessible at the trace level. Worked full-stack across the AI evaluation and observability platform using TypeScript, Next.js, PostgreSQL, and ClickHouse, building APIs, UI components, and data pipelines end-to-end. Resolved multiple production bugs, including a dataset re-import failure that blocked customers from running evaluators, traced via AWS CloudWatch to a soft-delete vs unique-key conflict and fixed by restoring deleted rows.