Thinking,Building & Engineering..HARSH BHARDWAJ

Architecting intelligent systems across the stack. From foundational models to deployed applications, building artifacts that prove capability.

03 / Worklog

Mission log, latest first.

Apr 2026 - Jul 2026

Contract Software Engineer

ClipHawk

Built ClipHawk as a contract-ready AI/data automation system designed to orchestrate agents, audit metadata, and scale high-volume entity processing.

Engineered a distributed orchestration server using Node.js and TypeScript to coordinate AI agents across worker nodes.
Scaled an event-driven data pipeline capable of processing 5,000+ complex entities daily without ingestion bottlenecks.
Integrated OpenAI and Anthropic APIs into production metadata-auditing layers, reducing manual review time per entity by 73%.
Implemented robust message queues for concurrent ingestion, reliability, and worker-level fault tolerance.

Node.jsTypeScriptAI AgentsOpenAIAnthropicMessage Queues

Jan 2026 - Apr 2026

AI Model Evaluation Contributor

Stealth

Worked on data annotation, model refinement, red-teaming, and guardrail evaluation workflows for state-of-the-art AI systems.

Contributed to annotation and evaluation workflows focused on model reasoning, safety, and instruction-following quality.
Used Docker-based environments to review, refine, and validate PRs related to model evaluation and red-team workflows.
Implemented and tested agent guardrails and evaluation flows across model behavior scenarios.
Maintained a cumulative quality score of 4.8 / 5 across assigned evaluation work.

AI EvaluationRed TeamingData AnnotationDockerGuardrailsModel Quality

Sep 2025 - Jan 2026

Cofounder

PlugInn

Cofounded PlugInn and worked across product, engineering, schema design, routing, GTM, market research, and founding strategy.

Designed core website flows, product schemas, and early platform architecture while contributing to founding product decisions.
Implemented location-based routing using Leaflet.js and built OTP token flows to initialize secure user sessions.
Built area-based caching to reduce repeated load on location-specific flows and improve platform responsiveness.
Led product research, market analysis, GTM definition, and product positioning through on-ground discovery.

StartupProduct StrategyLeaflet.jsOTP FlowCachingGTM

Jun 2025 - Sep 2025

Software Development Engineer

OneStopESG, Dubai

Worked on production-facing ESG platform features including jobs, authentication, API security, analytics, caching, and admin-side systems.

Built the ESG Jobs page end-to-end, contributing to a production surface that drew up to 80,000 users per month.
Worked on schema design and admin-side implementation for multiple endpoints alongside authentication flows.
Secured vulnerable Next.js APIs by implementing frontend token validation and strengthening request flows.
Implemented rate limiting, premium analytics, Redis caching, and NoSQL indexing for reliability and cost control.

Next.jsAuthenticationAPI SecurityRate LimitingRedisAnalytics

Jun 2024 - May 2025

Web Mentor

Google Developer Groups, MAIT

Started the development journey as a web mentor at GDG MAIT, contributing to the official website, mentoring students, and volunteering in Google's Gen AI learning cohort.

Worked on the official GDG MAIT website, contributing to web development and community-facing technical assets.
Conducted two student workshops focused on web development and practical technical learning.
Volunteered for the Gen AI Cohort 2025 and achieved Golden League recognition.

Web DevelopmentMentorshipWorkshopsGDGGen AI Cohort

// CODE ARCHIVE

Repository Atlas.

A monochrome proof map of the codebase: learning trails, small tools, ML experiments, archived context, and the projects worth inspecting.

Proof-of-Work Vault3 selected builds / click to inspect

Repository Atlas17 repos indexed

Contribution Frequencymapping activity

LessMore

Recent commitsactivity timeline

Jul 20267b9c3f1Created 1 commit in LearnOSlearnOSautomation trail

Jun 2026d4e1f8aCreated 10 commits across 4 reposNameFrame / WarWithin / ML101product + learning

Jun 20263c9a2b1Created 3 repositoriesWarWithin / ML101 / Personalnew build cycle

May 2026f1e7d9cOrganized repo proof bucketsportfolio maparchive work

Apr 20269b2d6e4Pushed screening and automation artifactsClipHawk-ScreenerML/data proof

Written Records

Field Notes

Systems, experiments, and the lessons left behind after the build.

01 / Field Note

How not to make Machine's Learn

A practical map of the places machine-learning systems break: data, sampling, features, evaluation, and assumptions.

May 29, 2026 4 reactions

Read dispatch

02 / Field Note

What the heck is IAM in AWS

A direct introduction to securing an AWS sandbox with identities, permissions, and deliberate access boundaries.

Aug 12, 2025 2 reactions

Read dispatch

03 / Field Note

Keep Your DB Awake with Hibernot

The build story behind a small automation that prevents free-tier databases from quietly hibernating.

May 25, 2025 7 reactions

Read dispatch

See all field notes

08 / Project Hitlist

Things I am trying to bring to life.

Ordered by what I want to attack next. Pick a priority to surface its world without turning the page into another card grid.

10 / About Me

Who am I ?

I'm Harsh Bhardwaj, an engineer from Delhi building full-stack systems, ML experiments, automation workflows, and public proof trails. I learn by turning ideas into visible artifacts: repos, notes, dashboards, notebooks, field records, and working tools that can be inspected.

01Observe02Build03Measure04Archive

Education

Maharaja Agrasen Institute of Technology

B.Tech CSE, 2023 - 2027 / CGPA 8.95 / ML & Data Analytics.

Study track

SIC Course in Data Analysis

Coursework focused on analysis concepts, structured data thinking, and practical interpretation.

Operating style

Useful systems over decorative builds

I prefer projects that leave proof behind: code, notes, metrics, automations, and decisions someone can inspect.