Mayflower

Leading a team of 12 MLE, DS, DA, DE, MLOps to improve:

1. Recommendation systems
2. Computer vision
3. NLP/LLM
4. Antifraud + Antibot + Antispam
5. Overall user behavior algorithms
6. DWH + overall Data and MLOps infrastructure

ML Team Lead

Was using NLP/LLMs/anomaly detection algorithms to end bot activities and help the Support & Moderation team on a website with 1B+ visits per month and hundreds of thousands of requests per day.

With a team of 2 in the Antispam department:

• Reduced the spam ratio from 1.4% to 0.13% (out of tens of millions of UGC content per day)
• Decreased the lifespan of a spammer on the website from 5.5 minutes to 12 seconds
• In a process, also reduced text moderation workload by 8.6x by lowering false positives

With a team of 18 in the Internal Tools department:

• Helped to adapt the internal tools to meet several government regulations (mostly in the EU)
• Transitioned most of the features used by internal teams in the old platforms to the new one with a better UX (saved 2 clicks per user per page on average)
• Had planned the addition of NLP/LLM agents and CV algorithms into ticket answering and content moderation, and continued building it in the new role with the ML team

Senior Technical Product Manager, AI/ML

LeadIQ

During my free time from a full-time job, I was giving consultations in prompt engineering, handling metrics, and growing the product that I owned previously

AI/ML Consultant

• Helped to launch an AI-powered freemium SaaS enabling SDRs to conduct prospect research and create personalized messages within seconds
• Achieved 21.5%+ growth for the like ratio of the messages
• Created a prompt engineering function, used a few LLMs (buzzwords: OpenAI GPT-3.5, GPT-4, Alpaca, Vicuna, GPT4ALL)
• Improved existing monitoring for messages performance (a/b testing, multi-armed bandits)

FUN BOX

• Researched and planned the creation of the VoiceBots platform for one of the largest mobile operators in the CIS region
• Hired a team of 1 developer and 1 data scientist, who helped create this platform from 0 to 1
• Scaled that team from 3 people to a whopping 5 when this platform started processing millions of phone calls per day
• Hired and mentored 2 junior PMs, they eventually started to lead 2 products in my portfolio (DMP and R&D + VoiceBots)
• Increased call duration (customer engagement metric) from 7 seconds to 39 seconds
• Helped to sell VoiceBots to a large telecom operator with 400% ROI.
• Established a process of creating new products that required a high level of R&D (new technologies) within a small budget

Group Product Manager, Data Science

Senior Product Manager, Data Science

Théa

• Established a Data Science function from scratch: product and hiring roadmap, quick wins, PnL, processes
• During the first few months was the only ML guy in the company which made me more an engineer than a PM
• Created data engineering pipelines that take raw data and insert it in Google BigQuery in a BI readable format with insights (geospatial analytics: Points of Interest detection, Map Matching functionality, distance calculation), all that with 140M data points per day in IoT-like hardware infrastructure (connected cars)
• Evaluated Domo, Looker, Tableau, Qlik for the BI platform for our business users to access the data and later picked Google Data Studio for us not to overspend on this. Created several dashboards in it
• For Ad-Hoc analysis wrote code that was putting our data into Valhalla+Kepler.gl for me to present it to our power users

• Took on an already successful A2P Platform and increased its revenue by 15%
• With a team of 5 MLEs created an antifraud functionality for one of the largest mobile operators in the CIS region
• Established a process for Data Science projects (ML/Deep Learning/MLOps) that later would become a company's standard
• With better product requirements created a way for a team of 25 engineers, data scientists, analysts and QA to communicate asynchronously almost with no meetings

Technical Product Manager

https://www.linkedin.com/redir/suspicious-page?url=VZV%2esu

• Saved a successful company from bankruptcy by noticing a huge business model "bug" during a financial crisis
• Led numerous client API integrations (Wildberries, Lamoda, Ozon - three of the largest e-commerce marketplaces in the CIS region)
• Designed a system that allowed us to easily parse pricelists of >50 suppliers and update availability and price of >15k SKUs daily and >5k SKUs every 15 minutes which was 3-4 times faster than the closest competitor
• To further automate the process of putting SKUs onto our website, wrote code for my first ML projects:

1. Footwear images classification (Keras+Inception v3)
2. Dominant colors in image detection (OpenCV+K-means)
3. A package's weight & volume prediction (DecisionTreeRegressor)

While CPO was on her maternity leave, I took full responsibility for managing a procurement & content team (~15 employees) and negotiating with ~50 suppliers in the largest children footwear distributor in Russia & CIS.

During that time I also was a main stakeholder for building our internal platform for a quick addition of goods on our website and importing suppliers’ pricelists which later took part in my transition to Product Management.

Deputy Chief Procurement Officer

During that job I was still a student at MPEI.

Here I polished my MS Excel skills and automated a lot of processes for fellow content managers via macros and formulas.

At the beginning I was happy to put 100 SKUs a day on our website, with my automation I was able to do more than a 1000 a day.

I still think of the ways to increase that number sometimes :)

Content Manager

Mustairs

My attempt at making something similar to Patreon or Buy me a coffee in 2012-2014.

This was my first entrepreneurial experience where I made lots of mistakes and I'm glad I did. There were no such products on the market: Patreon was founded in May 2013 and I found out about them a year later when they had steady growth. BMC was founded only in 2018.

This experience helped me to boost my product management skills at the time. I also based my final qualifying work in the MPEI on that startup which gave me an A score.

Founder

AI/ML Lead @ Mayflower

I'm a guy who takes on very complex products and improves them with AI (or simple algorithms where appropriate).

• Crafting a product vision? ✅
• Creating a data science department from scratch? ✅
• Advising on technical tools to use? ✅
• Coding an MVP? ✅
• Analyzing data stored in SQL and No-SQL databases? ✅ 
• Owning a PnL? ✅
• Hiring product managers and engineers? ✅
• Leading a team of 0 to 25 engineers, managers, and analysts? ✅
• Getting a product from 0 to 1 and making 400% ROI on it? ✅

Let's do things together!

--

9+ years in Product Management
7+ years in Data Science / ML / AI
2+ years in Prompt Engineering

🇨🇾 Limassol <- 🇷🇸 Belgrade <- 🇹🇷 Istanbul <- 🇹🇷 Antalya <- 🇷🇺 Moscow

I have a 🐕 Shiba Inu dog named Akio. Send a connection request for SHIB and DOGE luck!

Taplio

Sergei Zotov's Linkedin Analytics

Want to drive more opportunities from LinkedIn?

Taplio

.css-1jgreyo{background-image:linear-gradient(284deg, rgba(39,43,72,1) 0%, rgba(33,47,183,1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Sergei Zotov's Linkedin Analytics

Want to drive more opportunities from LinkedIn?

Sergei Zotov's Linkedin Analytics