AeroLeads people directory · profile
Ariel Tamches Email & Phone Number
Compute Performance
at
Lambda
LinkedIn matched
✓ Verified Jul 2026
3 data sources
Profile completeness 86%
Current company
Role
Compute Performance
Location
Los Gatos, California, United States
Company size
Who is Ariel Tamches? Overview
A concise factual answer block for searchers comparing this professional profile.
Quick answer
Ariel Tamches is listed as Compute Performance at Lambda, a with 580 employees, based in Los Gatos, California, United States. AeroLeads shows a matched LinkedIn profile for Ariel Tamches.
Ariel Tamches previously worked as Senior Staff Engineer at Weka and Senior Staff Engineer at Clumio. Ariel Tamches holds Ph. D., Computer Science from University Of Wisconsin-Madison.
Company email context
Email format at Lambda
This section adds company-level context without repeating Ariel Tamches's masked contact details.
Lambda
Review company-level records connected to Ariel Tamches before choosing the right outreach path.
Profile bio
About Ariel Tamches
Systems tech lead with 20+ years of experience, with expertise from machine code up to distributed systems, and all layers in between. I enjoy writing efficient, optimized, clean code and designs that bring together computer architecture, compilers, operating systems, data structures, and parallel lock-free algorithms. Proven on-time deliveries in both startup and post-startup companies. Multiple examples of redesigns with 10-100x performance improvement.Expertise in virtual machine hypervisors, block and file storage both on-prem and in the cloud, device driver development, low-level systems debugging, dynamic instrumentation, replication, resource management, and scheduling. System-wide performance analysis and optimization. Find and fix memory corruptions and diagnose CPU bugs. C++, C, golang, x86 assembly, gdb, parallel and lock-free algorithms. Improve performance, efficiency, test coverage, and maintainability of legacy code through successive optimizations, and cleanups, yielding a redesigned code base that is understandable from tech leads through to new graduates.
Current workplace
Ariel Tamches's current company
Company context helps verify the profile and gives searchers a useful next step.
Lambda
8 roles · 33 years
Ariel Tamches work experience
A career timeline built from the work history available for this profile.
Compute Performance
Senior Staff Engineer
Data plane: SaaS volume backup & restore, file and directory indexing & restore.Ownership and initiative in recognizing creative solutions to urgent need for AWS cost reduction. Architect of redesigns that fixed technical debt and yielded several orders of magnitude cloud cost savings and performance improvements.- Architect and Tech lead for Clumio NBD (network block device), allowing de-duplicated backups in S3 to be instantly and seamlessly accessed in a local EC2 volume. Used for file and directory level restores of customer volumes. By replacing redundant per-backup EBS volume copies that had been used for indexing and restore, this project reduced the company's highest component of AWS bills, making COGS sustainable.- Architect of NTFS optimized file indexing: highly optimized traversal of customer metadata. Using novel forensic analysis of low-level NTFS binary format, sped up directory and file indexing performance by over 90x, while improving AWS EC2 and S3 COGS efficiency by over 100x.- Architect and team lead for NBD Compaction project. By prefetching data blocks that are unchanged between backup generations, NBD gained another order of magnitude performance improvement and COGS reduction through reduced S3 and DynamoDB accesses, and reduced EC2 runtime. - Lock-free caching, profile-directed memory and CPU optimizations, scale-out performance improvements.
Aug 2019 - Oct 2023
Staff / Senior Staff Engineer
Early engineer (from stealth to post-IPO). Tech lead on Pure Storage FlashArray, from pre-beta to GA and beyond. Had the pleasure of working alongside some of the best engineers in a top-notch culture.Tech lead, resource management group. Scheduling, profile-guided optimization of memory performance (NUMA, lock-free allocators) and of CPU bottlenecks. Debugging memory corruptions and OOMs.Architect and co-author of FlashArray's completely lock-free highly parallel fairness task scheduler. Awarded 2014 Pure Performers Club for this work, which provided a unified solution to numerous blocker issues.Architect of FlashArray Real-time thread scheduler. Balancing Real-Time threads vs. background task threads to ensure responsive I/O along with smooth-running background activities. Novel mechanism for reducing the starvation caused by real-time threads through a measurement-based feedback controller. This work was awarded inaugural Pure Storage "Han Solo" award in 2015 from the Escalations/Support division, for solving urgent customer escalations quickly and elegantly.Memory allocator optimizations and tracking. Debugging of OOMs, memory leaks, and memory corruptions.Sync replication FlashArray: Optimized the performance of replicated small block writes by 10x (from 100 MB to 1 GB/sec) through benchmarking, profiling, code analysis, and bug fixes.Cloud Block Store pre-release data path engineer, specializing in optimization and debugging.Known as a company wide go-to person for toughest debugging issues, especially in solving memory corruptions.Discovered a pre-release CPU speculative execution bug by analyzing user-level core files; the bug had held up multiple product releases. Worked with FlashArray, DirectFlash, and FlashBlade teams to analyze the bug and coordinate diagnosis with Intel engineer and release managers. Part of engineering group who received company-wide award for outstanding achievement in this high-priority debugging effort.
May 2011 - Aug 2019
Software Engineer
Virtual Machine Monitor Development (core hypervisor). Worked with performance and vmkernel teams and their interaction with the VMM. Optimized vital code paths such as monitor to vmkernel calls. Wrote unified callstack code profiler, providing a novel correlation of performance spent amongst Guest OS, VMM, and VMkernel code. Low-level debugging, binary translation, x86 machine code. Worked with incredible teammates on a truly pioneering product.ESX 3.0 to 5.0, Workstation 4.5 to 7.0, Fusion 1.0 to 3.0. This work included the first virtualization of the 64-bit x86 instruction set, binary translation, hardware-assisted virtualization, multiprocessor VM support, and platform integration on Linux, Windows, Mac, and VMkernel host operating systems.Technical lead, ESX triple-vmx feature. ESX 5 ships with multiple versions of VMM and VMX: release, debug, and stats-gathering, and allows dynamic switching between the versions. This feature allows customers to run debug or profiling versions of VMware, and coordinate with technical support and developers to analyze the performance of the core hypervisor on customer systems.
Jul 2004 - May 2011
Software Engineer
Performance Lab and Volume Manager (VxVM) development. Worked directly with CTO and tech directors to identify performance bottlenecks in volume manager, dynamic multipathing, file system, remote replication, and clustering software.Designed and implemented numerous performance improvements to VERITAS Volume Manager (VxVM):- Using KernInst (my PhD project) to instrument VxVM kernel driver, identified L2 cache miss time bottleneck; added prefetch annotations to speed up RAID parity calculations by 400%.- Removed extraneous memory mapping and un-mapping occurring on the hot I/O path. Improved I/O initiation performance by 30%, and I/O completion processing by 40%—on the already-optimized RAID-0 code path.- Fixed VxVM buffer management bug that was causing large I/O’s on mirrored volumes to be broken up and dispatched as small chunks
Feb 2001 - Jul 2004
Phd Student
Advisor: Professor Bart MillerDissertation: Fine-Grained Dynamic Instrumentation of Commodity Operating System KernelsGraduate school web page, dissertation and other papers: www.cs.wisc.edu/~tamchesKernInst: Dynamic kernel instrumentation of commodity operating systems and its use in OS performance measurement and runtime optimizations. Parsing of raw OS machine code in memory into basic blocks (control flow graph), finding free scratch registers, at instruction granularity, available for use in runtime instrumentation (data flow analysis).
1994 - 2001
~7 yrs
3 education records
Ariel Tamches education
Ph. D., Computer Science
M.S., Computer Science
B.S., Computer Science
FAQ
Frequently asked questions about Ariel Tamches
Quick answers generated from the profile data available on this page.
What company does Ariel Tamches work for?
Ariel Tamches works for Lambda.
What is Ariel Tamches's role at Lambda?
Ariel Tamches is listed as Compute Performance at Lambda.
Where is Ariel Tamches based?
Ariel Tamches is based in Los Gatos, California, United States while working with Lambda.
What companies has Ariel Tamches worked for?
Ariel Tamches has worked for Lambda, Weka, Clumio, Pure Storage, and Vmware, Inc..
How can I contact Ariel Tamches?
You can use AeroLeads to view verified contact signals for Ariel Tamches at Lambda, including work email, phone, and LinkedIn data when available.
What schools did Ariel Tamches attend?
Ariel Tamches holds Ph. D., Computer Science from University Of Wisconsin-Madison.
Find 750M verified contacts
Search by job title, company, industry, location, and seniority. Export verified B2B contact data when you need it.
Start free trial