Harsh’s strategic insights and innovative AI solutions have made a real difference to our operational efficiency and campaign outcomes. His leadership consistently delivers meaningful and measurable results. Beyond that, Harsh is one of the kindest and most well meaning people you’ll meet, always looking to add value to whatever you’re building. He’s deeply intelligent, self aware, and thoughtful about the social impact of his work. A rare mix of sharp thinking and genuine intent, he is an extremely valuable asset to any team he works with.

- Sagar Gokhale, Co-founder, Qyuki

1. Insight Agent(2025)

Product: An AI-powered research and recommendation agent utilizing LangChain and Firecrawl for autonomous creator discovery.

Objective:

To automate and optimize the discovery of ideal creators by conducting comprehensive web research about brands and target campaign demographics, and generating a highly accurate creator list.

Problem:

Traditional research methods for creator selection were highly manual, time-intensive, and often lacked depth. Limited insights about brand compatibility and target audiences reduced campaign effectiveness and ROI.

Solution:

Integrated Insight Agent, an automated Retrieval-Augmented Generation (RAG) agent built using LangChain and Firecrawl. This AI agent autonomously searched and synthesized detailed online information about brands and targeted audience segments, providing an accurate and tailored shortlist of creators perfectly aligned with campaign objectives.

KPIs:

• Eliminated manual research efforts entirely (100% automation)

• Improved precision of creator-brand matches by 85%

• Accelerated campaign setup timelines by 70%

• Enhanced overall campaign ROI by delivering highly targeted creator partnerships

  
  
  

2. Qyuki Insights(2025)

Product: An AI-enabled data analytics tool designed to rapidly filter and identify ideal creators from an extensive global database.

Objective:

To simplify and accelerate the process of finding the most suitable content creators from a database exceeding 300 million entries, tailored precisely to specific campaign requirements.

Problem:

The team faced overwhelming data volumes, making the manual filtering of potential creators inefficient and ineffective. Identifying precise matches for brand campaigns required extensive time and often led to suboptimal choices, affecting campaign performance.

Solution:

Developed Qyuki Insights, an intelligent analytics platform leveraging advanced AI algorithms to swiftly analyze and filter creators based on highly detailed campaign criteria. The system rapidly narrowed down vast data sets to a targeted shortlist, significantly enhancing match accuracy for campaigns.

KPIs:

• Reduced creator identification time by 80%

• Improved campaign-to-creator alignment accuracy by 70%

• Increased campaign performance metrics (engagement rates, conversions) by 50%

• Enabled real-time adaptability in creator selection for dynamic campaigns

3. Qyuki Transcript (2024)

Product: An AI-powered video transcription and analytics platform leveraging Whisper APIs to extract actionable insights from multimedia content.

Objective:

To enable the creator management team to effectively analyze subtitles from YouTube videos to identify compelling hooks and highlights suitable for short-form video content.

Problem:

The creator management team struggled with manually reviewing subtitles across multiple YouTube channels to pinpoint engaging moments, chapters, and highlights, especially given language barriers and diverse content formats. This manual process was time-consuming, inconsistent, and often resulted in missed opportunities for high-impact short-form video production.

Solution:

Implemented Qyuki Transcript, an AI-driven transcription service powered by Whisper APIs. The system transcribed videos regardless of language, automatically identifying chapters, key highlights, and compelling hooks, streamlining content selection for short-form videos. This significantly reduced manual efforts, improved accuracy, and facilitated faster content creation decisions.

KPIs:

• Reduced video review time by 75%

• Achieved 95% transcription accuracy across multiple languages

• Increased content production efficiency by 60%

• Enhanced identification of viral content opportunities by 45%

  
  

  
  

4. Rekkoz Social-Commerce Platform

Social-commerce marketplace designed to facilitate influencer-driven sales in bandwidth-limited regions.

• Team Size: Directed an engineering organization of 15 across backend, frontend, UI, mobile, and DevOps teams.

  
  

Challenge: Delivering reliable, fast e-commerce functionality for influencer-driven purchases under challenging connectivity conditions.

Solution:

• Architected a modular, cloud-native infrastructure comprising 14 discrete microservices.

• Managed full-stack development of performant native mobile applications (Android, iOS) and a Next.js progressive web app.

• Ensured robust checkout processes, efficient inventory management, and real-time analytics.

  
  

Outcomes:

• Achieved 34,630 orders within the first month post-launch.

• Reliable sub-15-second checkouts, even on 2G connectivity.

• Rapidly scaled to over 310 sellers and 37,828 live SKUs.

5. Plutus Analytics Platform – JOSH Short-Video Ecosystem

Unified analytics platform architected to streamline insights for 500+ content creators.

• Client: Qyuki Digital Media

• Team Size: Led cross-functional team of 10 (backend, frontend, data engineers, DevOps).

  
  

Challenge: Fragmented analytics created significant delays (48-hour latency) in identifying viral trends and monetization opportunities.

Solution:

• Architected a scalable data lake using Apache Spark to centralize analytics.

• Directed development of interactive React/highcharts dashboards providing sub-minute insights into creator metrics.

• Implemented automated reporting pipelines, minimizing manual overhead.

  
  

Outcomes:

• Achieved a 90% reduction in reporting latency (from 48 hours to under 5 minutes).

• Enabled immediate responses to emerging trends, securing various sponsorship deals in the first month.

6. 2407 Creator-Commerce Control Tower

Unified dashboard solution for real-time analytics across multiple commerce platforms.

• Client: Qyuki Digital Media

• Team Size: Managed a dedicated agile squad of 4 engineers and analysts.

  
  

Challenge: Multi-store operations suffered from fragmented data, causing delayed insights and manual reconciliation efforts.

Solution:

• Designed and led the implementation of a centralized dashboard architecture using Azure Data Lake, Spark, and ELK stack.

• Established real-time data synchronization across Shopify, Amazon, and WhatsApp commerce channels.

• Automated crucial operational and financial insights, drastically reducing manual workloads.

  
  

Outcomes:

• Real-time revenue visibility achieved within a 15-minute refresh window.

• 95% reduction in daily manual reconciliation tasks.

• Enabled agile, daily budget reallocations instead of weekly.

  
  

7. Phi – Predictive Creator-Scoring Engine

Advanced predictive analytics engine for optimizing influencer discovery and onboarding.

• Client: Qyuki Digital Media

• Team Size: Supervised a multi-disciplinary team of 8 (data scientists, engineers, analysts).

  
  

Challenge: Inefficient, subjective influencer scouting resulted in delayed creator onboarding and missed opportunities.

Solution:

• Engineered a robust analytics platform leveraging Apache Spark to handle millions of data points nightly.

• Led the development of automated momentum scoring systems to rapidly identify potential breakout influencers.

• Implemented intuitive dashboards and alert mechanisms for talent teams.

  
  

Outcomes:

• Reduced time to identify and onboard high-potential creators by 53%.

• Increased successful high-growth signings by 28 percentage points.

• Enabled proactive sales engagement, resulting in exclusive sponsorship deals.