Krishna Anand · Software Engineer

I build production systems end to end.

Final-year Computer Engineering at MIT Manipal. Founding Engineer at Crescentia One LLP. Former SWE Intern at Bajaj Finance Ltd. Currently shipping production web and AI systems and prepping for SDE roles at top tech.

View work GitHub LinkedIn

LeetCode 1518
Top 4 of 2000+ at IIIT-H Felicity

Building

Founding Engineer at Crescentia One LLP

Shipping production web and AI systems end-to-end for retail and education clients. Storefronts, checkout, admin dashboards, and the deploy pipelines behind them.

Next.js
Express.js
PostgreSQL
Razorpay
AWS EC2

Learning

SDE prep and agentic systems

DSA and system design daily for SDE interviews at top tech. On the side, going deep on agentic LLM systems: orchestration, tool use, evals.

DSA
System design
LLM agents
Distributed systems

Selected work

Things I've built.

2026SWE Intern at Bajaj Finance Ltd

IT Request Advisory Agent

An advisory AI agent that reads database support tickets from Azure DevOps Boards and drafts risk-flagged SQL proposals for human review, without ever executing against a production database.

Built an advisory AI agent using Python and Azure OpenAI that reads database support tickets from Azure DevOps Boards, grounds them in a curated PostgreSQL knowledge base of 14 tables, and drafts risk-flagged SQL proposals for human review, with deterministic post-guards achieving 94.24% classification accuracy and zero false negatives on destructive SQL.
Designed a 7-stage pipeline (ingest, normalize, classify, resolve, generate, compose, notify) with three-stage classification and a four-tier SQL risk ladder that wraps destructive statements in rollback transactions and blocks DDL entirely.
Automated SLA metrics reporting via GitHub Actions and curated a 278-pair ground-truth dataset via the Azure DevOps REST API to quantitatively benchmark the agent.

Python
Azure OpenAI
PostgreSQL
Azure DevOps REST API
GitHub Actions

Case study

2026

WhatsApp PharmAgent

AI pharmacy assistant on WhatsApp with multi-pipeline orchestration over Azure

Master orchestrator on Azure Functions routes incoming WhatsApp messages between three pipelines, invoice OCR, voice billing, and natural language inventory queries.
Architected the data layer across 4 Cosmos DB containers (inventory, transactions, invoices, sessions) with phone-number partitioning and a session-based confirmation flow that survives function restarts.
Recognised in the Top 250 globally at Microsoft AI Unlocked 2026.

Python
Azure Functions
Azure OpenAI (GPT-4.1)
Cosmos DB
Document Intelligence
Speech-to-Text

Case study GitHub

2026Founding Engineer at Crescentia One LLP

Leopardol E-commerce Platform

Live at leopardol.com. End-to-end e-commerce platform for a Gen Z clothing brand, built as Founding Engineer at Crescentia One LLP

Owned the full backend (Express, Razorpay, Delhivery, support and email flows, rate limiting, cron) and led the engineering for a production e-commerce platform built from scratch.
Made the infrastructure calls, AWS EC2 for the Express backend and Cloudflare R2 over S3 for image storage (R2 has zero egress fees and lower per-GB pricing). Handled the deployment end to end.
Led client engagement and architecture decisions across the build.

Next.js
Express.js
TypeScript
PostgreSQL (Neon)
Prisma
Better Auth

Case study

2026Founding Engineer at Crescentia One LLP

ProfitMaster Website

Two-app financial-education platform for ProfitMaster (profitmaster.in), a Telugu-language stock market education company, built as Founding Engineer at Crescentia One LLP.

Built a two-app platform for a Telugu-language stock market education company, a public site on Cloudflare Workers at the edge and an admin dashboard on Vercel, both reading one Supabase Postgres schema.
Made concurrent seat reservation safe for capped-attendance seminars with a Postgres function that runs the capacity check and pending order insert in a single transaction, so a full event returns a clean sold out instead of failing at payment time.
Owned the user-facing site end to end (page routing, seminar registration, blog reading, Razorpay checkout with server-authoritative pricing), its deployment on Cloudflare Workers, and client engagement through the rollout.

Next.js
Cloudflare Workers
Vercel
Supabase Postgres
Razorpay
Cloudflare R2

Case study

2025SDE Intern at GoYaana (Aug to Oct 2025)

GoYaana Travel Chatbot

Conversational AI travel planner on AWS Lambda with LLM-driven itinerary generation

Built a serverless Java 21 backend on AWS Lambda that plans end-to-end trips from Bangalore to 118+ destinations, deployed via AWS SAM with API Gateway.
Used a dual-LLM strategy on Groq (Llama 3.3 70B for routing and date extraction, Llama 3.1 8B for batch description generation) to keep itinerary generation under the 29-second API Gateway timeout.
Containerised a Python Gradio frontend in Docker with FPDF2 and DejaVu fonts for Unicode and rupee support in downloadable PDF itineraries.

Java 21
AWS Lambda
AWS SAM
API Gateway
Groq (Llama 3.3 70B + Llama 3.1 8B)
Python

Case study

2026

TrackWell

Cross-platform mobile app for AI-powered nutrition tracking, fitness planning, and GPS-tracked activity

Three distinct Gemini AI integrations: vision-based food recognition from camera images, personalised daily fitness plan generation from onboarding inputs, and weekly health insights synthesised from 7 days of tracking data.
Real-time GPS walk tracker with foreground location streaming, Haversine-based distance calculation, and MET-based calorie estimation that uses the user's actual weight rather than a fixed multiplier.
Production touches across the app: 6-hour cache on weekly AI insights to keep API costs sane, debounced FatSecret OAuth 1.0 search, Clerk auth with Google OAuth, scheduled local notifications, and per-day log persistence in Firestore.

React Native (Expo SDK 54)
TypeScript
Expo Router
Google Gemini 3 Flash
Clerk
Firebase Firestore

Case study

2026

WorkflowPilot

Multi-agent system that automates enterprise workflows with autonomous error recovery and human-in-the-loop

LangGraph orchestrator routes between three sub-graphs (8-node onboarding, 5-node meeting, 5-node SLA) with conditional edges, retry-then-escalate error recovery, and human-in-the-loop pauses when LLM confidence drops below 0.5.
Dual-model LLM routing on Groq Llama 70B for reasoning and Gemini 2.0 Flash for extraction, with automatic fallback if the primary provider fails.
Immutable JSON Lines audit trail for every agent action, with reasoning, model used, and duration logged for compliance.

LangGraph
Groq (Llama 3.3 70B)
Google Gemini 2.0 Flash
FastAPI
React
Vite

GitHub

2025

MooMatch

AI-powered platform for indigenous Indian cow breed conservation and farmer support

Built the platform end to end: React frontend, Express backend (containerised on Cloud Run to keep the Gemini API key off the client), Flask ML backend, Firestore data layer, multilingual chatbot, marketplace, and dashboards.
Integrated an in-browser Google Teachable Machine model for real-time cow breed identification with no backend round-trip, alongside a Gemini-powered multilingual chatbot covering Hindi, Kannada, and Marathi.
The disease prediction Random Forest (trained on 90+ symptoms) was contributed by a teammate.

React
Tailwind CSS
Express.js
Flask
Firebase Firestore
Google Cloud Run

GitHub

Experience

Where I've worked.

Jun 2025 – Present

Founding Engineer

Crescentia One LLP

Production web platforms for retail and education clients.

Built and delivered Leopardol, a live retail e-commerce platform with Express.js + TypeScript backend, integrated authentication via Better Auth, Razorpay payments, and Delhivery shipping, on PostgreSQL with AWS EC2.
Shipped the ProfitMaster website, a live financial-education platform on Cloudflare Workers (user site) and Vercel (admin dashboard) with Supabase Postgres, integrated Razorpay payments and atomic seat reservation.
Chose Cloudflare Workers over Vercel and Cloudflare R2 over AWS S3 to keep pre-revenue infrastructure cost near zero; ran client meetings, owned architecture and timeline planning, and set PR-review and documentation standards.

Leopardol case study ProfitMaster case study

May 2026 – Jul 2026

Software Engineer Intern

Bajaj Finance Ltd

Advisory AI agent for internal IT request handling.

Built an advisory AI agent using Python and Azure OpenAI that reads database support tickets from Azure DevOps Boards, grounds them in a curated PostgreSQL knowledge base of 14 tables, and drafts risk-flagged SQL proposals for human review, with deterministic post-guards achieving 94.24% classification accuracy and zero false negatives on destructive SQL.
Designed a 7-stage pipeline (ingest, normalize, classify, resolve, generate, compose, notify) with three-stage classification and a four-tier SQL risk ladder that wraps destructive statements in rollback transactions and blocks DDL entirely.
Automated SLA metrics reporting via GitHub Actions and curated a 278-pair ground-truth dataset via the Azure DevOps REST API to quantitatively benchmark the agent.

Case study

Oct 2025 – Jul 2026

Director of Technology

MAHE MUN

Leading technology and web development for the university's Model UN organization.

Lead technology and engineering at MAHE MUN, overseeing web development, technical operations, and infrastructure.
Set technical direction for the MAHE MUN web platform, manage the developer team, drive code-review standards, and own architectural decisions across the org's web stack.
Built and shipped the official MAHE MUN 2025 website using React.js and Tailwind CSS, serving 400+ delegates with delegation portals, schedule management, and committee information pages.
Designed and implemented a dynamic newsletter system using the Google Docs API, allowing the content team to publish articles directly from Google Docs without developer intervention.

Aug 2025 – Oct 2025

Software Development Engineer Intern

Bestriptech Solutions Pvt. Ltd. (GoYaana)

Conversational AI travel chatbot on AWS Lambda.

Built a serverless Java 21 backend on AWS Lambda that plans end-to-end trips from Bangalore to 118+ destinations, deployed via AWS SAM with API Gateway.
Used a dual-LLM strategy on Groq (Llama 3.3 70B for routing and date extraction, Llama 3.1 8B for batch description generation) to keep itinerary generation under the 29-second API Gateway timeout.
Containerised a Python Gradio frontend in Docker with FPDF2 and DejaVu fonts for Unicode and rupee support in downloadable PDF itineraries.

Case study

Achievements

Hackathons and recognition.

4th of 2000+

IIIT-H Felicity Hackathon

2025Virtual Labs platform, 17M+ users

1st place

Kairotic Hackathon

2026Explainable misinformation detection

Top 250 globally

Microsoft AI Unlocked

2026WhatsApp PharmAgent project

Top 1500

Google Big Code

2026Through Qualifier, Round 1 of 15K, Round 2 of 1.5K

3rd place

Datavista

2025Built a model predicting GPS coordinates from real-time weather data.

Skills

Tools I work with.

Languages

Java
C++
Python
JavaScript
TypeScript
C
SQL

Frontend

React.js
Next.js
HTML5
CSS3
Tailwind CSS

Backend

Node.js
Express.js
FastAPI
Django
REST APIs

Databases

PostgreSQL
MongoDB
Firebase (Firestore)
Cosmos DB

Cloud & DevOps

AWS (Lambda, SAM, EC2)
Azure (Functions, OpenAI, Cosmos DB)
Google Cloud Run
Vercel
Netlify
Docker
Git
CI/CD

Get in touch

Let's talk.

Open to SDE internships and full-time roles. Cold outreach welcome.

krishnanand86654@gmail.com

linkedin.com/in/krishnaanand0

GitHub

github.com/krishna-086