Ayush Bhardwaj

BCA graduate (2026) who takes problems from definition to shipped, tested software, currently working full time as an AI and automation intern.

At The Elite

Point, manual research processes that used to take days now run in hours through LLM and workflow automation I designed and own. Comfortable across the full loop a product role needs: understand the problem, write the spec, pull the data, support the build, QA it, and document what shipped. Working knowledge of APIs, databases, and analytics; strong interest in LLMs, voice AI, and BFSI / SaaS platforms.

AI & Automation Intern at The Elite Point (2026-06 – Present)

• Cut a multi-day manual research cycle to a few hours. Owned the weekly lead-generation and intent-prospecting engine end to end (problem definition, workflow design, build, QA, rollout), replacing a manual multi-step process that consumed days of team effort each week.
• Shipped 20+ production automation workflows and AI agents in n8n used weekly by the team, each scoped with intended behaviour, edge cases, and defined failure handling (retries, token exhaustion, empty-result paths) before build.
• Designed a four-stage prospecting pipeline (discovery to exclusion to six-factor scoring to human review) with a master orchestrator running sub-workflows in parallel, backed by Supabase (PostgreSQL) for persistence and deduplication across profile ID, normalised URL, and name-company keys.
• Set the quality metric before shipping and checked it after: combined a deterministic 20-point ICP scoring model with LLM classification (Fit / Unsure / Not a fit), then compared output against reviewer decisions and tuned thresholds, raising qualified-lead accuracy to 80% and cutting manual review volume by 60-70%.
• Continuously look for the next bottleneck to automate. Iterate on the scraping and enrichment layer each cycle to raise throughput and cut runtime (from 700 to 3,000 records per week), and bring the team proposals for what to automate next based on where time is still spent manually.
• Built custom scrapers for HTTP basic-auth, authenticated sessions, and JavaScript-rendered pages that auto-paginate and export structured XLSX, backed by an Apify token pool with rotation, exhaustion handling, and weekly key rollover that keeps runs continuous at near-zero cost.
• Automated reporting and delivery: per-segment Excel reports with formula-injection guarding, dispatched via Gmail, with every batch logged to the database so any output traces back to its run.

Full Stack Developer Intern at Avijo Healthcare (2026-03 – 2026-05)

• Found and closed a critical security flaw on the live site where OTPs were exposed in network responses; documented the issue, patched the email-OTP authentication flow, and eliminated a high-severity production risk.
• Resolved 15+ cross-module UI and API defects across patient profiles, medical records, payments, subscriptions, uploads, and PDF generation, each reproduced, written up with expected vs actual behaviour, and verified before release.
• Restored account-management flows reported by real users: email verification, password change, linked Google/Facebook accounts, privacy controls (deactivation/deletion), and display settings.
• Hardened the React.js frontend and MongoDB/Mongoose backend against undefined-data edge cases, aligned API contracts, and tightened query logic and schema validation to prevent runtime failures.

Bachelor of Computer Applications in BCA – Dehradun Institute of Technology University (2023-08 – 2026-08)