Solution Architect (Python, Go)

For one of our clients, we are looking for a Solution Architect - Python & Go (m/f/n).

Offer

• 100% remote
• Rate: up to 195 PLN/h net+VAT
• B2B contract via SHIMI
• Start: 01.10

Core languages

• Go: concurrent services, UDP/TCP servers, high-throughput decode loops, profiling with pprof.
• Async Python: asyncio internals (event loop, tasks, cancellation, backpressure), async HTTP client/server, multiprocess worker pools.

Quality discipline

• Static typing (mypy / golangci-lint), formatters/linters, pre-commit gates.
• Async unit tests, snapshot tests for decoders, load tests in CI.
• Dependency/CVE hygiene as a reflex, reproducible pinned builds.

Networking depth

• L2/L3/L4 headers by hand. VLANs, MAC/IP semantics, TCP flags, MTU/fragmentation, IPv4 + IPv6.
• Flow telemetry protocols: IPFIX (RFC 7011), NetFlow v5/v9, sFlow v5. Template-based decode, sampling rates, sequence gaps and loss handling.
• Packet capture and crafting; reading pcap as a primary debugging tool.
• Network device automation over SSH/CLI and vendor APIs (Cisco, Arista, Juniper, Palo Alto). Idempotent config push, drift detection.

Data movement

• Protocol Buffers: schema design, evolution, codegen. Fast JSON/binary codecs for hot paths.
• Kafka producer/consumer semantics: partitioning, acks, batching, compression, SASL/Kerberos, lag diagnosis. NSQ/NATS a plus.
• Columnar/timeseries sinks (ClickHouse or similar).

Performance engineering

• Profiling as routine, not incident response. Allocation and GC pressure, lock contention, zero-copy paths.
• Right data structures: LPM/Patricia tries for subnet lookup, LRU caches, fast hashing. Can explain the tradeoff, not just name it.
• Load generation and sustained-throughput testing.

Systems and delivery

• Linux internals: sockets, epoll, sysctl tuning for high packet rates, cgroups/resource limits.
• eBPF a strong plus.
• Containers + Kubernetes: multi-stage builds, fast inner-loop dev, manifests.
• Shipping a self-contained binary to customer infrastructure: static/frozen builds, deb/rpm packaging, upgrade and rollback paths.
• IPC patterns (ZeroMQ or equivalent).

Distributed state

• Cache/state-sync design, eventual consistency, reconnect-and-resync after long disconnects, idempotent replay.

Identity and cloud integrations (plus)

• Windows/AD plumbing: WinRM, SMB, Kerberos.
• A major cloud SDK (Azure or AWS) for asset and log ingestion.