Datadog Interview Questions — Page 2

Datadog·Machine Learning Engineer·Technical Phone Screen

May 2026

Technical screen for an ML Engineer role at Datadog focused entirely on implementing Grouped-Query Attention from scratch. Pretty deep dive into transformer internals, not the usual LeetCode grind.

Implement the forward pass of a Grouped-Query Attention module from scratch using only basic tensor operations, no high-level attention helpers.
How much memory does GQA save versus standard multi-head attention during autoregressive decoding, and why is memory bandwidth the bottleneck at decode time rather than raw compute?
How would you convert an existing multi-head attention checkpoint into GQA? Specifically, how do you initialize the G KV heads from the original H KV heads?
Where do rotary position embeddings get applied in a GQA setup, and does the grouped structure change anything about that?
Implement the incremental decode step for GQA: given a cached K and V of length t and one new token, produce the next output and update the cache. What shapes change?

“The core of the whole interview.”

View Post

Datadog·Software Engineer·Onsite - System Design / Architecture

May 2026

Datadog system design round, one big question about building a log ingestion and query system from scratch. Pretty intense for a single question but they really dug into the details across the full hour.

Design a data structure that records log entries and supports efficient querying. Each log has a timestamp in milliseconds, a severity level, a service ID, and a message. You need to support: appending a log, counting logs in a time range with optional filters, and fetching logs in a time range with filters, ordering, and pagination. Walk through your data layout, indexing strategy, time-range scan approach, and the complexity tradeoffs between write amplification and query latency.

“I started with a sorted structure keyed on timestamp, which felt right, but then they pushed on what happens when you add severity and serviceId filters on top of the range scan and I kind of fumbled.”

View Post

Datadog·Software Engineer·Onsite - System Design / Architecture

May 2026

Did a system design round at Datadog for a software engineer role. Just the one question, pretty open-ended, not a lot of context given upfront.

Design a system to find a specific photo within a continuous data stream.

“There's a lot of directions you can take this and I think that's kind of the point.”

View Post

Datadog·Software Engineer·Onsite - Behavioral / Leadership

May 2026

Interviewed for an EM role at Datadog, got a behavioral question about people problems. Pretty standard leadership screen but the question had more depth to it than I expected.

Tell me about a people-related problem you've invested significant time and energy into.

“I had a story ready but fumbled the ending a bit.”

View Post

Datadog·Software Engineer·Technical Phone Screen

May 2026

Datadog software engineer interview with a coding question about bucketing sorted integers. Pretty straightforward problem but there are a couple of edge cases worth thinking through.

Given a sorted array of integers, a bucket width, and a number of buckets, distribute the values into buckets and return the count per bucket. The last bucket is a catch-all for anything that overflows the regular range.

“The core logic is pretty clean once you see it: for each value just do integer division by the bucket width and clamp to the last bucket index.”

View Post

Datadog·Frontend Engineer·Onsite - System Design / Architecture

Jun 2026

Datadog system design round with two interviewers, no specific system given upfront. You're expected to drive the whole thing yourself, from requirements through architecture to optimization, and the pacing pressure is real.

Design a system in the Datadog domain (e.g. metrics ingestion pipeline, time-series storage, dashboarding, or alerting). Requirements are not given upfront; you are expected to gather them yourself and drive the entire session.
How would you handle time-series data compression and downsampling at scale in a metrics storage system?
Walk through how you would optimize the system you designed. What are the bottlenecks and how would you address them?

“Two interviewers, no prompt beyond the general domain.”

View Post

Datadog·Software Engineer·Technical Phone Screen

Jun 2026

Datadog SWE interview with two back-to-back coding problems, both leaning more system-design-ish than pure leetcode. The questions were practical enough that I didn't feel totally lost, but the complexity analysis follow-ups were rougher than I expected.

Design a log store that accepts entries with ISO-8601 timestamps and unique IDs. Implement add(id, timestamp) and a query(start, end, granularity) method where granularity can be year, month, day, hour, minute, or second. Walk through your data structures, time and space complexity, and how you handle ordering and memory.
Implement a BufferedWriter that wraps an abstract sink whose write(bytes) method may only accept partial writes. Your implementation should support write(data), flush(), and close(). How do you handle ordering, chunking large inputs into fixed-size buffers, minimizing system calls, partial write errors, and thread safety?

“I went with a sorted structure keyed on parsed timestamps and figured that was enough.”

View Post

Datadog·Software Engineer·Technical Phone Screen

Jun 2026

Datadog software engineer interview that centered on a fairly meaty design-and-implement problem around building a query client for a data warehouse. The follow-ups pushed into security and batch processing territory, which I wasn't fully prepped for.

Design and implement a lightweight query client for a Snowflake-like data warehouse. It should support connecting via runtime config, submitting async SQL queries, polling for status, retrieving results, and handling errors like invalid SQL, missing credentials, network failures, and unknown query IDs.
How would you handle secrets like database credentials securely in this client?
How would you extend the client to support submitting multiple queries in batch and polling their statuses together?

“This one took me a minute to scope.”

View Post

Datadog·Machine Learning Engineer·Technical Phone Screen

Jun 2026

Datadog ML engineer interview threw a pretty gnarly systems question at me about fitting OLS on a dataset that won't fit in memory. The whole thing felt more like a design discussion than a coding screen, which I wasn't fully prepared for.

You have a 25 GB CSV file with over a million rows and 3,000 float features per row. It doesn't fit in memory. How do you fit an OLS regression on it? Walk through streaming aggregation, numerical stability, parallelization, and how you'd verify your result against a small baseline.

“This one took me a minute to even parse.”

View Post

Datadog·Software Engineer·Onsite - System Design / Architecture

Jun 2026

System design round at Datadog for a software engineering role, basically a full YouTube design from scratch. Pretty broad scope and they pushed into a lot of subsystems, so it was a lot to cover in one session.

Design a video platform like YouTube, covering upload, transcoding, storage, streaming, and user-facing features like comments, likes, and recommendations.

“This one sprawls fast.”

View Post