CV.pdf ↓
Senior Python backend engineer · Saint Petersburg · UTC+3

Andrey
Vasiliev

I build distributed backends that stay fast under load: ingestion at 40M+ records, analytics in under 1.5 seconds, semantic search in under 400 ms.

4+ years in production · FastAPI, Django, PostgreSQL, ClickHouse, RAGSaint Petersburg · UTC+3 · 4+ years

Get in touch → github.com/N3ffus ↗
~/andrey — zsh80×24
$ whoami
andrey vasiliev — senior python backend
$ cat stack.txt
python · fastapi · django · postgresql
clickhouse · redis · celery · kafka · pgvector
$ leetcode --solved
1,000+ problems ● accepted
$
Query latency
✕ 30–60s →
<1.5s
OLAP reports moved to ClickHouseOLAP on ClickHouse
API response
✕ ~8s →
<250ms
N+1 fixes, indexes, Redis cacheindexes + Redis
Records ingestedRecords
20K+ messages / day20K+ / day
40M+
with transactional consistencyingested
Vector retrievalRetrieval
50K+ documents50K+ docs
<400ms
● 99.9% availability99.9% up
01 / Selected work

Systems in production

Strikt · 2024 – now · featured

Real-time Telegram analytics & ingestion engine

Distributed ingestion for 20K+ messages a day and 40M+ stored records. Reporting moved from PostgreSQL to ClickHouse materialized tables; the API sits behind Redis with fixed N+1 queries and rebuilt composite indexes.

✕ 30–60s →
<1.5s
analytical queries, −95%
✕ ~8s →
<250ms
API response, −90%
Telegram20K+ msg/day
Celery workersingest · dedupe
PostgreSQLOLTP · 40M+ rows
ClickHouse MVhot path · <1.5s
API + Redis<250ms p50
Telegram › Celery › PostgreSQL › ClickHouse MV › API + Redis
PostgreSQLClickHouseRedisCelery
Strikt · AI platform

Enterprise knowledge & semantic search

<400ms
vector retrieval over 50K+ docs

End-to-end RAG: embedding pipelines, pgvector retrieval and multi-agent workflows on DeepSeek and Qwen, with checks against hallucinated answers. 99.9% availability.

FastAPIpgvectorDeepSeekQwen
Strikt · Vedomosti events

Conference management & payments

10K+
attendees across 25+ conferences

Zero-downtime services with 30+ REST endpoints under strict OpenAPI contracts and idempotent YooKassa payments. Test coverage raised to 80%+, regressions down 40%.

DjangoDRFYooKassaPytest
Freelance · 2022 – 2024

Job aggregation engine

20K+
listings a day, 99% dedup precision

Scrapers and ETL across multiple job boards with backpressure and retries — 30% fewer data anomalies. aiogram bots deliver updates to 2,000+ daily subscribers.

FastAPIRabbitMQCeleryaiogram
02 / Experience

Where I've worked

09/2024 — now
Remote

Senior Python Backend Engineer · Strikt

  • Built ingestion for 20K+ daily messages and 40M+ records with strict transactional consistency.
  • Cut analytical queries from 30–60s to under 1.5s by moving OLAP workloads to ClickHouse.
  • Cut API response time from ~8s to under 250ms: N+1 fixes, composite indexes, Redis.
  • Mentor 5+ backend developers; ran 20+ technical interviews.
  • Introduced Clean Architecture, ADRs and shared CI/CD templates — new services onboard in half the time.
09/2022 — 08/2024
Remote

Python Backend Engineer · Freelance

  • Built job aggregation processing 20K+ listings a day with 99% deduplication precision.
  • Designed ETL on FastAPI, Celery, Redis and RabbitMQ with backpressure — 30% fewer data anomalies.
  • Shipped aiogram bots and notifications for 2,000+ daily active subscribers.
03 / Stack

What I work with

■ daily in production □ used in projects
01 / Languages & core~/stack/core [6]
  • Python
  • SQL
  • AsyncIO
  • Algorithms
  • 1,000+ LeetCode
  • Codeforces
02 / Backend~/stack/backend [11]
  • FastAPI
  • Django
  • DRF
  • Pydantic
  • SQLAlchemy
  • Alembic
  • REST
  • OpenAPI
  • Microservices
  • Clean Architecture
  • SOLID
03 / Data & storage~/stack/data [10]
  • PostgreSQL
  • ClickHouse
  • Redis
  • pgvector
  • MongoDB
  • S3
  • Neo4j
  • Qdrant
  • Query optimization
  • Sharding & replication
04 / Messaging~/stack/queues [8]
  • Celery
  • RabbitMQ
  • Kafka
  • FastStream
  • Dramatiq
  • Event-driven
  • Task queues
  • Background jobs
05 / AI & LLM~/stack/ai [10]
  • RAG
  • Embeddings
  • Semantic search
  • AI agents
  • LangGraph
  • Haystack
  • OpenAI API
  • DeepSeek
  • Qwen
  • DeepEval
06 / ML & NLP~/stack/ml [9]
  • PyTorch
  • Transformers
  • sentence-transformers
  • SetFit
  • GLiNER
  • scikit-learn
  • ONNX Runtime
  • MLflow
  • pandas
07 / Scraping & bots~/stack/scraping [6]
  • Playwright
  • Patchright
  • Selenium
  • BeautifulSoup
  • Telethon
  • aiogram
08 / Testing & quality~/stack/quality [6]
  • Pytest
  • Hypothesis
  • mypy
  • Ruff
  • pre-commit
  • coverage.py
09 / DevOps~/stack/devops [12]
  • Docker
  • Kubernetes
  • GitLab CI/CD
  • GitHub Actions
  • Nginx
  • Caddy
  • Linux
  • Prometheus
  • Grafana
  • Sentry
  • OpenTelemetry
  • Yandex Cloud
04 / Contact

Let's talk backend

Distributed systems, data pipelines, AI infrastructure. Telegram is the fastest way to reach me.

Write on Telegram ↗
$ contact --telegramt.me/neffus ↗
$ contact --email$ andrew.vasilev03@gmail.com
$ contact --github$ github.com/N3ffus ↗
$ contact --leetcode$ leetcode.com/u/N3ffus ↗
$ languages$ Russian — native · English — B2