Hi, I'm Abhishek Kumar

AI Systems Engineer • Backend Engineer

Building production-grade AI systems that solve real problems. Specialized in agentic AI, multi-agent orchestration, and autonomous copilots.

View Projects ↓ Download Resume

About Me

I'm a software engineer with 3+ years of hands-on experience designing and shipping production-grade AI systems. My journey started in full-stack development, but I found my passion in building intelligent systems that genuinely solve problems.

At Kindlife, I've architected end-to-end AI products deployed across multiple organizations. From personal AI copilots that autonomously handle emails and meetings, to company-wide systems managing purchase orders—I own systems completely, from architecture through deployment.

What drives me is the challenge of making AI systems that don't just work, but work reliably at scale. Whether it's orchestrating multiple agents, building episodic memory systems, or optimizing infrastructure for 1,500+ daily conversations, I'm focused on bridging the gap between research and production reality.

Years Building AI Systems

1,500+

Daily Conversations Handled

Organizations Deployed To

Tech Stack

AI/ML

Agentic AIMulti-Agent OrchestrationRAGEpisodic MemoryVector SearchPrompt EngineeringLLM Tool UseMCP (Model Context Protocol)LangChainLangGraphLangFuseHuman-in-the-LoopSupervisor-Worker Patterns

Databases

PostgreSQLMySQLNeo4j (Graph DB)Qdrant (Vector DB)BigQuery

Languages & Frameworks

PythonDjangoFastAPINode.jsPHPJavaSQL

Infrastructure & Tools

KafkaTemporalDockerGCPAWSGitSentence Transformersn8nRundeck

My Journey

Software Engineer

Kindlife

Apr 2022 – Present

▸Architected Personal & Company-Wide AI Copilot deployed across 3 client organizations
▸Designed Skills system enabling non-technical users to automate repetitive tasks
▸Built customer support bot handling 1,500+ conversations daily (50% reduction in human intervention)
▸Implemented multi-agent orchestration with supervisor-worker patterns
▸Built episodic memory pipelines using Neo4j and Qdrant
▸Eliminated ~600 redundant SQL queries, reducing response time by 50%

Full-Stack Intern

ShopClues

Apr 2021 – Apr 2022

▸Integrated AI-powered recommendation engine driving 17.6% of platform revenue
▸Optimized frontend performance (PageSpeed from <10% to >90%)
▸Implemented WhatsApp transactional messaging pipeline for order notifications

Education: B.E. Computer Science • Chitkara University, Chandigarh • 2022 • 9.77 CGPA

Problems I've Solved

Real-world challenges in production environments

Personal & Company-Wide AI Copilot

Problem:

Automating complex workflows across enterprises with AI

Solution:

Built an autonomous copilot system that reads/sends emails, schedules meetings, queries live databases, generates insights, and executes ERP actions via natural language.

Impact:

Deployed across 3 organizations, handling complete automation workflows

Multi-database integration (BigQuery, MySQL, PostgreSQL)Real-time email & meeting automationERP system integrationSkills system for non-technical users

Intelligent Customer Support Bot

Problem:

Handling 1,500+ daily support conversations efficiently

Solution:

Engineered a production GenAI agent with adaptive filter discovery and self-learning capabilities for product recommendations.

Impact:

50% reduction in human intervention, scalable semantic retrieval with live product context

1,500+ daily conversationsDomain-specific search engine with adaptive filtersReal-time embedding pipelinesVector database integration (Qdrant)

Performance & Infrastructure Optimization

Problem:

Application bottlenecks and slow response times at scale

Solution:

Profiled application, eliminated 600 redundant SQL queries, implemented query caching layer, and built scalable notification system.

Impact:

50% reduction in end-to-end response time, capable of reaching hundreds of thousands of users

Query optimization via ProxySQL cachingCRM push notification systemFirebase Cloud Messaging integrationReal-time delivery tracking

AI-Powered Recommendation Engine

Problem:

Driving revenue growth through personalized product recommendations

Solution:

Integrated machine learning recommendation engine that analyzes customer interaction history to suggest relevant products.

Impact:

17.6% of platform revenue driven by AI recommendations

Personalization algorithmCustomer behavior analysisReal-time recommendationsRevenue impact optimization

Let's Connect

Interested in discussing AI systems, opportunities, or just saying hi? Reach out!

Send Email Call Me LinkedIn

abhi98166@gmail.com

+91 8091736331

Chandigarh, India