Vishist Pandey profile photo 1 on Tal

Vishist Pandey

CTO at Qwikgig

Bengaluru

3+ years of work experience

Sep 20251 yPresent

Bengaluru

Building Auditsail
QwikGig
Sep 20251 yr 1 moPresent

Founding/sole architect on a small startup team - hands-on across AI, backend, frontend, and infra; title reflects startup scope, day-to-day work is IC engineering. Initialized and architected the entire Auditsail platform from scratch: a multi-tenant SaaS where organizations can invite and share campaigns with other organizations, cascade access to internal and external members, with an integrated credit/billing system layered on top. Built on polyglot persistence to match each domain's access pattern: PostgreSQL on AWS RDS (primary + read replicas) for core/org/billing data, MongoDB for AI output and an immutable audit trail, and Redis as an event bus syncing independently-schemaed services in real time. Architected an event-driven AI pipeline (AWS S3 → SQS → transcription → compliance & script-adherence analysis → MongoDB) that has processed 1.2M+ call recordings, end-to-end in ~20 seconds for an average 5-minute call. Implemented multi-provider LLM routing (OpenAI, Google Gemini, Mistral, AWS Bedrock), selecting the lowest-cost model that meets accuracy targets per task, holding inference cost to ~$0.04/audio-minute; added response caching for a further 18% cost reduction. Built an LLM evaluation workflow with Langfuse against a golden-labeled, compliance-officer-verified dataset, reaching 93.47% compliance-detection accuracy with a repeatable loop for prompt/model improvement. Solved a scale bottleneck by indexing the call-recordings listing table in Elasticsearch after Postgres query latency degraded to 3-5 seconds at million-row scale, cutting queries to milliseconds; built a Redis-backed real-time dashboard and Grafana monitoring (latency/throughput, error rates) for compliance teams, client organizations, and on-call visibility. Evaluated build-vs-buy on transcription: migrated from self-hosted Whisper + pyannote to AssemblyAI after self-hosting hit rate-limit ceilings and infrastructure cost/management overhead.
QwikGig
Mar 20257 mosSep 2025

Built and led a high-performing AI team of 3 members, responsible for full-cycle delivery of intelligent automation projects. Designed and deployed a scalable architecture to ingest call recordings from international call centres, transcribe audio, perform compliance checks, and auto-integrate results into the company’s CRM with near-zero human involvement. Developed and implemented AI-powered pipelines that streamlined production deployment, reducing operational delays and manual workload by 90%. Created an LLM-powered internal chatbot integrated into the CRM dashboard, offering real-time stakeholder insights and analytics from operational data. Successfully implemented OCR functionalities within the CRM to autonomously extract, process, and integrate structured data from unstructured documents. Collaborated with development teams to identify and prototype automation opportunities using AI, increasing process efficiency across business units.
Anagram Media Labs
Sep 20246 mosFeb 2025

Delhi, India

• Designed algorithms to automate data scraping from diverse sources, scaling the database from 0 to 4.5M records in just 6 months. • Built efficient pipelines for data collection and storage, improving operational workflows by 70%. • Recruited and led a 5-member tech team in a fast-paced startup environment, delivering projects on tight deadlines. • Trained interns, enhancing team productivity by 35% through mentorship and optimized task delegation. • Developed predictive models to analyze real-time data streams, enabling data-driven strategies and insights. • Built and managed data pipelines for Instagram data scraping, resulting in actionable business insights. • Partnered with the DevOps engineer to implement CI/CD pipelines, reducing deployment times by 40%. • Containerized data pipelines and ML models for efficient, scalable deployment. • Acted as a bridge between stakeholders and the technical team, presenting challenges and solutions effectively to drive consensus and decision-making.
Ubamarket Ltd
Feb 20247 mosAug 2024

London Area, United Kingdom

• Developed data pipelines using Big Query, improving processing efficiency by 25%. • Tracked key metrics (basket size, basket duration, common products), enhancing data accuracy by 30%. • Utilized advanced analytical techniques, increasing actionable insights by 20%. • Integrated LLM APIs to develop a predictive model for product recommendations, boosting sales growth by 18%. • Improved app features with LLM APIs, resulting in a 15% increase in usage. • Collaborated with cross-functional teams, reducing decision-making time by 10%. • Contributed to data-driven strategies, increasing overall business efficiency by 12%.
Vodafone
Jun 202311 mosApr 2024

Stratford-Upon-Avon, England, United Kingdom

• Surpassed sales targets by 150% monthly, consistently meeting all KPIs. • Delivered exceptional customer service, building strong relationships and maintaining high client satisfaction. • Demonstrated extensive product knowledge to provide tailored solutions, increasing business growth by 25%. • Trained team members in sales skills, enhancing overall department performance by 30%. • Used Oracle CRM software for account management, improving efficiency. • Created and interpreted store and employee performance dashboards using Tableau, boosting business efficiency by 18%. • Collaborated with stakeholders and contributed to strategy building, driving business growth.

Vishist’s AI Stack

claude code
claude code
claude
claude
codex
codex
cursor
cursor
kiro
kiro
claudecode
claudecode

Education

Coventry University

Masters, Data science

Coventry University

20221 yr2023
Guru Gobind Singh Indraprastha University

Bachelor of Technology - BTech, Mechanical and Automation

Guru Gobind Singh Indraprastha University

20183 yrs2021

Join Vishist and 10k+ other professionals on

Tal Boss

Tal is the app where bosses meet engineers & builders directly, without middlemen

Download Tal Boss

Vishist Pandey on Tal

Headline

CTO at Qwikgig

About

CTO at Qwikgig

Location

Bengaluru

AI stack

claude code

claude

codex

cursor

kiro

claudecode

Work experience

Chief Technology Officer at Auditsail, Sep 2025 - Present · 1 y, Bengaluru, Building Auditsail

Lead Engineer at QwikGig, Sep 2025 - Present · 1 yr 1 mo, Founding/sole architect on a small startup team - hands-on across AI, backend, frontend, and infra; title reflects startup scope, day-to-day work is IC engineering. Initialized and architected the entire Auditsail platform from scratch: a multi-tenant SaaS where organizations can invite and share campaigns with other organizations, cascade access to internal and external members, with an integrated credit/billing system layered on top. Built on polyglot persistence to match each domain's access pattern: PostgreSQL on AWS RDS (primary + read replicas) for core/org/billing data, MongoDB for AI output and an immutable audit trail, and Redis as an event bus syncing independently-schemaed services in real time. Architected an event-driven AI pipeline (AWS S3 → SQS → transcription → compliance & script-adherence analysis → MongoDB) that has processed 1.2M+ call recordings, end-to-end in ~20 seconds for an average 5-minute call. Implemented multi-provider LLM routing (OpenAI, Google Gemini, Mistral, AWS Bedrock), selecting the lowest-cost model that meets accuracy targets per task, holding inference cost to ~$0.04/audio-minute; added response caching for a further 18% cost reduction. Built an LLM evaluation workflow with Langfuse against a golden-labeled, compliance-officer-verified dataset, reaching 93.47% compliance-detection accuracy with a repeatable loop for prompt/model improvement. Solved a scale bottleneck by indexing the call-recordings listing table in Elasticsearch after Postgres query latency degraded to 3-5 seconds at million-row scale, cutting queries to milliseconds; built a Redis-backed real-time dashboard and Grafana monitoring (latency/throughput, error rates) for compliance teams, client organizations, and on-call visibility. Evaluated build-vs-buy on transcription: migrated from self-hosted Whisper + pyannote to AssemblyAI after self-hosting hit rate-limit ceilings and infrastructure cost/management overhead.

Artificial Intelligence Engineer at QwikGig, Mar 2025 - Sep 2025 · 7 mos, Built and led a high-performing AI team of 3 members, responsible for full-cycle delivery of intelligent automation projects. Designed and deployed a scalable architecture to ingest call recordings from international call centres, transcribe audio, perform compliance checks, and auto-integrate results into the company’s CRM with near-zero human involvement. Developed and implemented AI-powered pipelines that streamlined production deployment, reducing operational delays and manual workload by 90%. Created an LLM-powered internal chatbot integrated into the CRM dashboard, offering real-time stakeholder insights and analytics from operational data. Successfully implemented OCR functionalities within the CRM to autonomously extract, process, and integrate structured data from unstructured documents. Collaborated with development teams to identify and prototype automation opportunities using AI, increasing process efficiency across business units.

Data Scientist at Anagram Media Labs, Sep 2024 - Feb 2025 · 6 mos, Delhi, India, • Designed algorithms to automate data scraping from diverse sources, scaling the database from 0 to 4.5M records in just 6 months. • Built efficient pipelines for data collection and storage, improving operational workflows by 70%. • Recruited and led a 5-member tech team in a fast-paced startup environment, delivering projects on tight deadlines. • Trained interns, enhancing team productivity by 35% through mentorship and optimized task delegation. • Developed predictive models to analyze real-time data streams, enabling data-driven strategies and insights. • Built and managed data pipelines for Instagram data scraping, resulting in actionable business insights. • Partnered with the DevOps engineer to implement CI/CD pipelines, reducing deployment times by 40%. • Containerized data pipelines and ML models for efficient, scalable deployment. • Acted as a bridge between stakeholders and the technical team, presenting challenges and solutions effectively to drive consensus and decision-making.

Data Scientist at Ubamarket Ltd, Feb 2024 - Aug 2024 · 7 mos, London Area, United Kingdom, • Developed data pipelines using Big Query, improving processing efficiency by 25%. • Tracked key metrics (basket size, basket duration, common products), enhancing data accuracy by 30%. • Utilized advanced analytical techniques, increasing actionable insights by 20%. • Integrated LLM APIs to develop a predictive model for product recommendations, boosting sales growth by 18%. • Improved app features with LLM APIs, resulting in a 15% increase in usage. • Collaborated with cross-functional teams, reducing decision-making time by 10%. • Contributed to data-driven strategies, increasing overall business efficiency by 12%.

Sales Advisor/Analyst at Vodafone, Jun 2023 - Apr 2024 · 11 mos, Stratford-Upon-Avon, England, United Kingdom, • Surpassed sales targets by 150% monthly, consistently meeting all KPIs. • Delivered exceptional customer service, building strong relationships and maintaining high client satisfaction. • Demonstrated extensive product knowledge to provide tailored solutions, increasing business growth by 25%. • Trained team members in sales skills, enhancing overall department performance by 30%. • Used Oracle CRM software for account management, improving efficiency. • Created and interpreted store and employee performance dashboards using Tableau, boosting business efficiency by 18%. • Collaborated with stakeholders and contributed to strategy building, driving business growth.

Education

Masters, Data science, Coventry University

Bachelor of Technology - BTech, Mechanical and Automation, Guru Gobind Singh Indraprastha University

Public links

vishistpandey.com: https://vishistpandey.com/