|
Vladimir PreobrazhenskiyDeveloper / Tech Lead / AI Engineer
|
💬 About me
IT specialist with over a decade of experience in developing and
managing innovative products.
Passionate about leveraging AI to create impactful solutions that drive business value and
client's
satisfaction. Expert in conversational AI driven systems, Architecture, DevOps practices, and
cloud
infrastructures, with a strong background in programming and data analysis.
Proven leader with the ability to join or manage cross-functional agile teams to deliver top
solutions
and
products. Eager to collaborate with forward-thinking companies to invent the future of
technology
and
the
way we
live.
Russian (native),
English (B2).
🔸 Conversational AI: ~5 years in the field + Won RASA 2025 AGENT BUILDING CHALLENGE. Applied LLM Prompting, finetuning.
🛠️ Core skills
📅 Career path
🔷 StoryJar
| Period | 2024.02 - 2024.12 |
| Role | Hybrid Role: 50% Product Owner, 50% Architect/DevOps |
| About |
StoryJar is a name of POC activity, aimed to test several hypothesises: - Heavily prompted modern LLMs can produce high quality stories on user's behalf. - Conversational capabilities, being added to story generation AI agent provide smoother UX and engage user much more. - Such stories are essential part of family media archive representation, driven by smart AI agents in miltimodal UX environment. |
| Tech stack |
DEV/INFRA:
Github (with actions),
Terraform,
AWS,
Digital Ocean,
DockerARCHITECTURE: NGINX,
Postgres,
RabbitMQ,
S3DEV: Python,
Fastapi,
Insightface,
GRPC,
websockets,
JWT auth,
Flutter,
Figma
Miro
|
| Responsibilities |
- Clarifying requirements with stakeholders. - Detailing the research agenda. - Identifying key features of proposed prototype. - Architectural decomposition of prototype. Business architecture, Technical Architecture. - Planning and conduction particular steps of RnD process. - Budget management. - Team management. - Team guidance on cornerstone topics. - Regular reporting on POC status. - Attending strategic sessions with stakeholders. - Maintain CI/CD process. |
|
|
| Key Projects & Achievements |
✅ Family Archive Prototype 💪 Hired team, set up end 2 end development process based on kanban: 💪 Special product Achievements: 🔹 Transparent async media files upload and streaming. 🔹 Automated media description process. 🔹 Semantic search across media objects. 🔹 Custom style and design. 🔹 Can be installed locally or in any cloud. 🔹 Realtime voice input for comments and descriptions. 🔹 Automated Face detection and face recogition for user's images. accuracy ~97%. At strategic session in the end of 2024 prototype was successfully presented to investor. We revised our approach and agreed to switch team's focus to story generation capabilities. Therefore, prototype was frozen and Writer AI Agent phase started. ✅ Writer AI Agent Work In Progress here, having several goals: - Enable automated story generation on top of Prototype's backend. + Add conversational features for engaging UX - Identify criteria for good story. - Tune story generation mechanism. - Invite 10-30 test user's and collect feedback via questionnaire. Additionally: - Add face recognition for user's videos. - Add scene description to user's videos. - Enhance Writer to be able to work with various document types. |
🔷 QRit
| Period | 2023.07 - 2024.02 |
| Role | DevOps |
| About | QRit is a platform designed to help restaurants and hospitality businesses offer a seamless, multilingual, and cashless experience for their customers using QR codes. |
| Tech stack |
DEV/INFRA:
Github (with actions),
Terraform,
Digital Ocean,
GoDaddy,
DockerARCHITECTURE: NGINX,
Postgres,
Redis,
S3DEV: Figma,
Node + TypeScript,
Next + TypeScript,
|
| Responsibilities |
- Maintain CI/CD Pipelines. - Cloud Infrastructure management. - Database architecture. - Techical analysis for 3rd party integrations. |
|
|
|
| Key Projects & Achievements |
✅ QRit 💪 Set up cloud infrastructure with Terraform + GoDaddy + Digital Ocean. 💪 Configured CI/CD pipelines for TEST, PROD stands. 💪 Provided Database ER model for DEV team. 💪 Provided Streamlit based Admin panel for early stage of product. 💪 Provided Streamlit based Admin panel for early stage of product. |
🔷 Insitech Development
| Period | 2021.05 - 2023.12 |
| Role | Tech Lead |
| About | Realtalk was established to develop a multiuser software complex that enables goal-oriented voice dialogues across a wide range of scenarios. |
| Tech stack |
DEV/INFRA:
Gitlab (with CI),
Ansible,
Terraform,
Docker SwarmARCHITECTURE: Microservices,
SIP,
Yandex Cloud,
NGINX,
Postgres,
Redis,
DEV: Python,
BERT based custom intent detection model,
Vue + TypeScript
|
| Responsibilities |
- Interview and hire tech candidates. - Set up and maintain DEV + delivery process. - Architecture design. - Infrastructure management. - POC of new features. - Participation in preparations for tenders |
|
|
|
| Key Projects & Achievements |
✅ Realtalk project After my experience in X5 I was invited to lead a small team who was aimed to provide similar services to the market. I had to recompose team for best results: 💪 Reached top competitors convertion rate for NPS/CS collection case in 1 year after product development started. 💪 Delivered core component, that handles telephone conversation in realtime with features: 🔹 Speech Syhthesis / Predefined recordings. 🔹 Detection of multiple intents from peer's utterance. 🔹 Interruption requests handling by bot. 🔹 Dialogue creation framework with web based GUI. 🔹 Scalability. Up to 1M call attempts per day. 🔹 Improved stability, maintainability and computational efficiency, compared to the previous solution. 🔹 Friendly GUI provided for campaign management, reporting and dialogue scenario creation. 🔹 Digressions handling (FAQ, silence, chitchat attemps, etc..) 💪 Realtalk has been added to the official Russian software registry. 🔹 I had to completely redesign technical architecture and delivery process to make the solution meet the requirements. 🔹 Achieved higher level of fault-tolerance as a bonus. 💪 Having 50+ cloud servers under control for DEV/TEST/UAT/PROD stands, I've automated infrastructure management with terraform + ansible 💪 Developed inhouse stands management system to ease delivery process. |
🔷 X5 Retail Group
| Period | 2017.11 - 2021.04 |
| Role | Hybrid Role: 50% Tech Lead/Developer, 50% Analyst/PM |
| About | X5 Retail Group is Russian leader in grocery retail. |
| Tech stack |
INFRA:
Gitlab (on-prem installation),
K8S (on-prem installation)
DEV/ARCH: Scala,
SIP,
Python
|
| Responsibilities |
- Participation in product shaping. - Running POC for internal clients. - Conducting interviews with tech candidates. - Develop solution within enterprise arch landscape. |
|
|
|
| Key Projects & Achievements |
✅ RnD in Video Analytics and Speech Technologies 💪 In close interaction with the management of the newly formed Big Data department, participated in the preparation of a large RnD case. Case being defended consists of two major areas (Video Analytics and Speech Technologies). I've: - (1) Initiated this activity and gained support from my chief. - (2) Prepared business case to be defended on C level of X5 Group. - (3) Took part in the defense of the case before the company's C-level management. Being interested in NLP related activities, I switched my focus to Speech Technologies completely. 💪 As a developer (later tech lead), developed and implemented a system that allowed conducting dialogues with retail chain customers over the phone when they contacted the call center. The entire technology stack was developed and implemented within the organization, with minimal involvement of contracting organizations. 💪 Defened inhouse solution to be embedded into Enterprise's arhitecture landscape. 💪 The flagship scenario implemented on the basis of the developed solution was the registration/replacement of a loyalty card by voice without involving people, with one of the best conversion rates on the market. (1) ~10000 incoming calls on a daily basis, ~6500 callers asked for loyalty card registration/replacement. (2) 70% of inquiries are being solved without redirection to call center. 💪 Developed and implemented a customer support bot in the WhatsApp channel. 💪 Expanded tech team (up to 5 Dev/DevOps/DS positions), ensured the implementation of an outbound calling system for collecting NPS + CS. 💪 Added services responsible for coordinating outbound traffic in accordance with the organization's contact policies. Solution is able to handle several hundreds thousands calls per day. ✅ Point of Sale Inventory Management 💪 Prepared a data showcase on product stock balances for 5 subsidiary. Revealed and identified stocks for several billion of rubles. Ensured smooth knowledge transfer to a dedicated team continued with the internal IT product. |
🔷 Beeline (Big Data Department)
| Period | 2015.09 - 2017.04 |
| Role | Hybrid Role: 50% Dev/DS, 50% PM/Analyst |
| About | Beeline is a well known russian federal level digital provider. Company owns an industrial Hadoop cluster with petabytes of varying-quality data. The division was focused on delivering business value using open-source on-premise Big Data infrastructure. |
| Tech stack |
DEV/ARCH:
Hadoop,
SPARK,
SQL,
Python,
Java,
Shell
|
| Responsibilities |
- Conduct division's activities. - Implement ML based solutions. - Adopt best practices in Project/Product workflow delivery process. - Integrate advanced technologies into daily operations. |
|
|
|
| Key Projects & Achievements |
✅ Division's workflow reorganization 💪 Manager: Identification and analysis of current business processes to find inefficiencies. 💪 Manager: Adopting agile practices to ongoing activities with the support of specialists from the headquarters in Amsterdam. ✅ Client's gender identification. Reached 85+% accuracy 💪 Analyst/DS: Identified the most influential predictive features. Prepared Specifications for ETL process. 💪 DS: Trained logit model for gender prediction using SPARK. 💪 Manager: With the help of a contractor, implemented and automated ETL for generating input data to the model. 💪 Here I was first in the Big Data department who delivered production-grade SPARK-based solution. ✅ Churn prediction 💪 Data Engineer: Implemented SPARK based ETL pipeline that generates features for churn detection model. ✅ Customer's golden record 💪 Manager: With contractor's help, delivered daily ETL process that collects various information about Beeline's customers. |
🔷 Accenture
| Period | 2012.03 - 2015.08 |
| Role | Hybrid Role: 70% Dev/DevOps, 30% analysis (requirements, validation) |
| About | Global IT consulting leader. website . |
| Tech stack |
DEV:
Python,
Java,
Shell,
SQL,
Informatica,
Teradata,
Hadoop,
Anylogic,
Photoscan,
Linux
|
| Responsibilities |
- Deliver modeling, analytics, and data processing pipelines. - Provide scalable, fault-tolerant solutions. Ensure integration of diverse components. - Participate in internal R&D and customer-facing projects. - Prepare documentation and specs. |
|
|
|
| Key Projects & Achievements |
✅ Automated Infrastructure Inspection via Drones (R&D Initiative) 💪 Client Engagement: Conducted on-site requirements gathering with energy sector clients. Coordinated drone flight operations with 3rd-party partners (aircraft-type UAVs, 1000km range) 💪 DevOps: Managed aerial imagery processing pipeline (Photoscan, Linux). Worked with CV engineers to deploy algorithms detecting infrastructure defects. ✅ Social Media Lead Generation (Top Russian Bank) 💪 Developer: Crawled 50+MLN VK users with their core attrubites and custom features representing their ability to influence 💪 DevOps: Built 10-node Hadoop cluster to process crawled data. ✅ E-Commerce Warehouse Simulation 💪 Analyst/Developer: In collabortaion with retail experts designed and implemented Anylogic model that cut order assembly time by 30%. ✅ Risk Data Mart for Banking Reserves 💪 Analyst: Collection and clarification of requirements. 💪 Developer: Implementation of ETL process with Informatica + Teradata. |
🔷 Luxoft
| Period | 2009.11 - 2012.02 |
| Role | Junior Developer |
| About | Luxoft is a multinational technology company providing software development services and IT solutions. |
| Tech stack |
DEV:
C,
Java,
Shell
|
| Responsibilities |
- Participate in agile rutuals (Scrum, Planning, Grooming, etc..). - Write tests, create and review code. Bug fixing. |
|
|
|
| Key Projects & Achievements |
✅ ZigBee stack implementation for semiconductor manufacturers 💪 Consolidated knowledge in C, Java. 💪 Created ZigBee protocol stack simulator, accelerating test cycles. 💪 Gained low-level programming experience working with boards/microcontrollers. |
🎓 Education
Higher
Higher School of Economics, Department of Sociology,
specialization
"Analysis of Social Institutions". 2010, Bachelor.
Defended a graduation thesis on the topic "Constructing the Mechanism of Trust towards Machines
as
Subjects
of
Social Relations".
Additional
- In 2024.10. Kubernetes - Complete Course on Kubernetes. Certificate Link
- In 2018. Training at Skolkovo. Трансформация и рост бизнеса в эпоху новых технологий (Business Transformation and Growth in the Age of New Technologies)
- In 2016. Beeline Data School (9 month course with wide range of ML related topics). Beeline.
- In 2016. (1) Agile SW development, (2) Requirement Management in Agile projects. ScrumTrek
- In 2013. Mastering Innovation & Design Thinking. MIT. Cloudera:
- In 2013. Introduction to Data Science: Building Recommender Systems. Cloudera.
- In 2013. Cloudera Certified Specialist in Apache Hbase. Cloudera.
- In 2012. Cloudera Certified Developer For Apache Hadoop. Cloudera.
- In 2012. Cloudera Certified Administrator for Apache Hadoop. Cloudera. Basic:
- In 2010. Java, Java EE. "Specialist" at Bauman Moscow State Technical University.
- In 2009. Visual C/C++. "Specialist" at Bauman Moscow State Technical University.