Every project grounded in a real problem, a real approach, and a real outcome.
9+9 scoring signals · 75+ priority threshold · 24hr contact SLA · Close rate: 52% → 71.63%
Only 10% of leads are ready to buy. Built the B2B GTM framework that identifies which 10% — scoring prospects on 9 intent signals and 9 potential signals before a single outreach message is sent.
17:1 adoption gap · 19:1 E-Invoice gap · C-Suite escalation · −58% Rejection rate
Conducted deep-dive analysis uncovering a massive drop-off where PRO+ users generated E-Way Bills externally via Tally rather than in-app. Built an executive narrative that changed the product roadmap.
Checkout abandonment: 73.1% → 53.9% (–26%) · 480K sessions · Beat Wayfair on mobile CVR
Ran a friction audit combining GA4 step funnels and Clarity session evidence. Simplified fields, clarified trust signals, and adjusted error/validation UX.
Session-to-PDP-click: +17% · Qualified leads: +34% · Bounce: −16% · ATC: +27%
Led a 4-week sprint guided by GA4 custom events and Microsoft Clarity heatmaps. Replaced commodity category pages with story-driven, trust-led journeys. 1.1M BigQuery events analyzed.
Lead submissions: +124% · Mobile completion time: −41% · Rage clicks: −68%
Diagnosed event funnels and rage-clicks. Rebuilt static form into category-specific modules with Material 3 components, contextual microcopy, and latency fixes.
Gold source CVR: 63.5% · Bronze source CVR: 0.4% · Days to close: 5.2 → 3.5 target
Built a data-backed lead routing system across 4 agents. Website forms (63.5% CVR) routed differently from chat (4.7% CVR). 30-day pilot from 10% to full rollout.
+168% feature engagement · 27 occasions · 5 geo-regions
Integrated Google Gemini Flash (Nano Banana) image AI into LiveKeeping's dormant Pro+ Send Greetings feature. Built a geo-segmented festival calendar across 5 Indian regions with 27 occasions.
27+ triggers · P0–P3 priority · 3-slot daily cap · 5 geo-regions
Redesigned the entire lifecycle messaging architecture covering renewals, transactional states, conflict logic, and feature releases across PRO and PRO+ plan segments.
3 sources unified · 88 rows mapped · 11 AM auto-populate · 0 manual steps
Built a Google Apps Script pipeline unifying Kibana, MongoDB, and GA4 into a single auto-populated report. Eliminated manual data entry across the team.
478 sign-ups/month · +51% from baseline
Redesigned the full-funnel webinar experience — landing page, email sequence, and post-session nurture. Mapped drop-off points across registration, attendance, and follow-up before rebuilding each touchpoint from the data up.
Cost-to-book reduced by 36%
Transitioned from manual LinkedIn prospecting to a structured outbound system with lead scoring, BDR capacity planning, and a tiered qualification framework based on company size, vertical, and AI readiness signals.
16-page whitepaper · 12 verticals mapped
Produced a 16-page research report covering the agentic AI tooling landscape, market sizing, adoption maturity curve, and 8+ enterprise case studies across manufacturing, legal, logistics, and finance verticals.
Revenue model · Pre-seed fundraising support
Built the revenue architecture and tiered pricing model for Upcore's AI agent services. Partnered with the CEO on financial modelling and the investor narrative as part of pre-seed fundraising preparation.
20+ interviews · Top-3 opportunity brief delivered
Conducted structured discovery interviews with 20+ prospective clients across 12 verticals. Synthesised findings into a prioritised opportunity brief identifying the top-3 agentic AI use cases by severity, ROI, and buildability.
19:1 E-Invoice adoption gap identified and reported to C-suite
Investigated low utilization of the native E-Invoice module, mapping workflow disconnects between LiveKeeping and Tally's default integrations using the same methodology as the E-Way Bill diagnosis.
Weekly CEO dashboard tracking GST rejection errors · Reduced manual reporting effort
Engineered a multi-week tracking instrument summarizing complex GST compliance rejection codes across product tiers. Created a standardized data pipeline for executive visibility.
Bounce rate: −22% · AOV: +8%
Used GA4 funnels and Clarity scroll/click insights to ship speed-tuned, modular landing templates that match paid-traffic intent.
Custom-order attach rate: +28%
Defined real-time quote logic and shipped a lightweight JS widget embedded on PDPs with validation and analytics hooks.
Search-attributed revenue: +14%
Re-weighted relevance factors, added synonym mapping, and ran query-intent experiments with controlled A/B rollout.
Support tickets: −40% · NPS: +6 points · Cross-sell revenue: +43%
Defined requirements, selected vendor, and designed track-my-order and self-serve flows with instrumentation. Reduced support load and unlocked cross-sell opportunities.
Add-to-cart rate: +27% · Page engagement time: +22%
Event-driven iterations on media gallery, variants, and micro-copy, informed by GA4 custom events and Clarity behavior data.
Products I design, code, and ship myself — with the repo public. Every line, commit, and backtest visible.
Precision retirement planning for Indian investors
SIP, EPF, PPF, NPS, and FIRE calculators with inflation-adjusted, institutional-grade projections.
Self-hosted LinkedIn outreach agent
You describe your target roles; the agent finds the right recruiters and hiring managers, writes personalised AI intros, and manages follow-ups. One-click macOS app. 766 commits, 6 releases.
Bollywood greenlight risk engine
A 10-component scoring engine blending gradient-boosted ML with Bayesian priors, trained on 2,200+ Indian films — with brutally honest published backtests and known failure modes.
Procurement intelligence for manufacturers
Explodes sales orders through BOMs, nets demand against supply, computes min stock per component, and back-schedules POs — with trapped working capital quantified in rupees. 28 reconciliation tests.
SEO content pipeline for non-technical teams
Pulls real GSC data to find content revival and gap opportunities, drafts with Gemini (fact-check pass included), creates WordPress drafts — never auto-publishes — and logs every run to Sheets.
Voice AI that runs entirely in your browser
Named after Feluda’s assistant in the Ray stories — the one handed the research and planning — Topshe runs wake word, speech-to-text, and a quantised Qwen 2.5 model in-browser over WebAssembly. No server, no API bill, no data sent anywhere. Offline-capable PWA at $0/month.
Movie night, solved for couples
Picks your next film together based on mood — built for exactly two people. Live on Vercel and in active nightly service at home.