Open to opportunities

Harshita Oberoi

@harshitaoberoi

Message

Software Engineer specializing in AI/ML, Fullstack, and DevOps—delivering low-latency, accurate systems.

India

Message

What I'm looking for

I’m looking for a team where I can own AI/ML to production pipelines—improving latency and accuracy—while shipping fullstack features and maintaining robust CI/CD and cloud infrastructure.

I’m a results-driven Software Engineer specializing in AI/ML, fullstack, and DevOps. In production-style builds, I delivered a 30% reduction in trading latency, achieved 95% forecasting accuracy in healthcare AI, and scaled APIs to 5K+ DAUs.

I build cloud-native, distributed systems that bridge ML models, microservices, and modern web technologies. From agentic AI workflows and multi-agent pipelines to secure REST APIs and dashboards, I automate deployments and cut operational overhead—like saving 4–5 hours/day through workflow automation—while keeping latency, accuracy, and reliability front and center.

Experience

Work history, roles, and key accomplishments

Backend developer

Solytics Partners

Mar 2026 - May 2026 (2 months)

Built internal automation tools using Python (Flask) and Google Sheets API.
Automated reporting workflows for 10+ business teams.
Processed and transformed 1M+ rows using Pandas and SQL.
Built REST APIs following OOP and Low-Level Design (LLD) principles.
Used Git for version control and production deployments.

Remote Flask Django REST APIs Google Analytics Google Sheets Git GitHub Actions GitHub

Software Engineer Intern

Krins Digital

Engineered a crypto futures trading bot using RSI/EMA/MACD, improving win rate by 15% and reducing latency by 30%. Built secure REST APIs and React/Next.js dashboards for real-time analytics, and automated CI/CD with Docker and GitHub Actions to deploy to AWS EC2 with zero downtime.

Crypto Trading Systems EMA REST APIs fastAPI React Next.js Docker GitHub Actions

Software Engineer Intern

Goldman Sachs

Built agentic AI workflows with LangGraph/LangChain (Router + Supervisor) and scaled multi-agent pipelines for 10K–15K row datasets via data chunking. Applied CoT, ReAct, and Reflexion to reduce hallucinations and improve accuracy, and automated processes saving 4–5 hours/day.

LangGraph Lang Chain Multi Agent Systems Chain Of Thought React Reflexion Workflow Automation