Built at Apple and Instacart.
Now we build with you.

AIshar Labs started with a question I kept hearing from founders: "Where do I find someone who can actually build this?"

I spent nearly a decade inside Apple and Instacart — building the ML systems behind search, recommendations, and personalization that hundreds of millions of people use every day. At Apple, I built & powered search for Maps, Safari, and Spotlight. At Instacart, I architected the ML platform for recipe recommendations, product search, and feed ranking. I filed 15 AI patents because we were solving problems nobody had solved before.

But here's what I noticed, over and over: the companies outside big tech couldn't access this level of engineering. They'd hire consultancies that delivered strategy decks but couldn't ship production models. They'd outsource to agencies that built prototypes that collapsed under real traffic. They'd spend months trying to recruit ML talent that Apple and Google had already locked up.

The gap wasn't knowledge.
It was engineering depth.

Founders and enterprise teams had the domain expertise. They understood their markets, their users, their data. What they didn't have was someone who'd built ML systems at billion-query scale and could apply those patterns to their specific problem.

That's the gap AIshar Labs fills. We're not a consultancy that learned AI. We're AI engineers who spent years building at the highest level — and now bring that same rigor to every company we work with.

We don't advise. We embed with your team and build. We don't create dependency. We transfer everything and make sure you can run it without us. And we don't staff projects with juniors. The architect who designs your system is the same person who writes the code.

AIshar Labs is deliberately small, intentionally senior, and completely focused on one thing: building production AI systems that work.

15+ years of building ML at scale.

Apple

2016 – 2021 · Lead ML Engineer

Search: Maps, Safari, Spotlight

Built on-device personalization and phrase-level spelling correction for Apple Search. Shipped multiple ML and deep learning models under strict latency and privacy constraints. Multiple innovations showcased at WWDC (2019-2021).

~5yr

tenure

3+

patents

B+

queries

Instacart

2021 – 2025 · Staff ML Architect

Search, Recs, Ranking, Autocomplete

Architected recipe recommendation, product search, feed ranking, and autocomplete ML systems. Designed embedding-based retrieval and contextual bandit frameworks. Primary author on 10+ patents. Multi-million dollar revenue impact.

~4.5yr

tenure

10+

patents

$MM+

impact

Adobe

Adobe

2011 – 2014 · Sr. Technical Staff

Cloud Services, Analytics, Reader

Led data and ML initiatives across Adobe's creative and document workflows. Drove improvements to Reader's Create PDF and Export PDF features, plus cloud-based analytics within massive distributed systems.

3yr

tenure

Sr.

staff

CIIR UMass Amherst

CIIR, UMass Amherst

2014 – 2016 · Research

Machine Learning, Information Retrieval & NLP Research

Published research at ACM on query reformulation and word sense disambiguation. Center for Intelligent Information Retrieval — one of the world's leading IR research labs. MS in Computer Science.

2

publications

MS

CS degree

The proof behind the work.

📚

15

AI patents filed (personalization, retrieval, embeddings, ranking)

🏆

8th

ACM ICPC Asia (international competitive programming)

🔬

100%

TBDC Mentor · Antler Alumni · Top conference publications & research

🎯

G

Google for Startups Accelerator Americas Alumni

Deliberately small. Intentionally senior.

We don't scale with headcount.
We scale with depth.

AIshar Labs is a small team of senior AI engineers — not a bench of consultants waiting for their next project. Every person on our team has built production ML systems. Nobody is learning on your project.

We stay small on purpose. It means the person who designs your architecture is the same person who writes the code, reviews the models, and stands behind the system in production. No handoffs. No communication gaps. No dilution of quality.

When a project requires specialized depth we don't have, we bring in people we've worked with before — not strangers from a staffing bench.

  • 🛠

    Architects who code

    Every team member designs and builds. No one is "strategy only."

  • 🎯

    Production experience required

    We only hire people who've shipped ML to real users at real scale.

  • 👥

    No bench staffing

    We don't maintain a bench. Your team is chosen for your specific problem.

  • 🤝

    Embedded, not outsourced

    We work inside your standups, repos, and Slack. Not in a separate silo.

Google for Startups

Accelerator Americas

Antler

Antler

Entrepreneur in Residence

BHive

BHive

EIR & AI Mentor

TBDC

TBDC

AI Mentor

Publications and patents — not blog posts.

Academic research and patented inventions, not recycled frameworks.

Patent

Personalized Recommendation of Recipes Based on Embeddings

Method for personalized recommendation using embeddings for users and stored recipes.

View patent →

Patent

Personalization in Instant Search

System for personalizing instant search results using ML models under strict latency constraints.

View patent →

Publication · ACM 2016

Automatic Iterative Reformulation of Queries with Aspects until Convergence

Published at ACM CIKM. Research on automatic query understanding and reformulation.

View publication →

Publication

Probabilistic Latent Semantic Analysis for Unsupervised Word Sense Disambiguation

Research on probabilistic methods for understanding word meaning in context.

View on Google Scholar →

Now you know who we are.
Tell us what you're building.

A 30-minute conversation with our founder. No pitch deck. Just an honest discussion about your problem and whether we're the right team to solve it.

Talk to Manmeet